I was stuck in a spiral where I pretended to myself and others that I was attracted by them in order to receive attention, affection, emotional support, care, understanding.

Male socialization probably made me have a way harder time believing anyone could be happy for me to open up with them and be my entire self if they were not a partner.

I think that’s the thing I was missing in a bunch of cases, the need to be my entire self.

Becoming very in love and infatuated was some sort of unintentional love bombing, so people would not be tempted to leave me.

Oh, shit that hurts to think of my behavior as emotional bombing, but it’s not so far. I do think I genuinely cared about all the people I’ve been partnered with, even when I pretended to myself to care more than actually did.

I remember Vals telling me that me having many relationships was a bad sign. It showed that I did not know how to end relationships. I ended multiple ones after that. Including one which was definitely limping along not from attraction and love, but for pity towards my partner, from fear of what’s they’d become without me.

I’m not sure what’s been the driver of those times of intense infatuation which always end up decreasing.

One of the drivers of this dynamic is that a new relationship had no baggage. If I’m not capable of processing the stress and conflicts of the relationship, they accumulate, and the relationship becomes less and less good, and interests ceases. A new relationship has no baggage, so it’s new and shiny and simple.

But that was the easy one, the dynamic I’ve known for a long time.

One that I don’t really know how to point at yet is one around something that I was looking for in those relationships, that somehow I did not find, and stopped believing I could find, which made me lose interest. And the high level of excitement at the start came from feeling like getting this thing was possible and just within reach, but it never really materialized

Possibly it is some sort of chasing dreams of a relationship where I would just be so obviously compatible with the person, that I would not have to do any effort to make it go well? Hm, not sure if that’s a main part.

I think a thing that kept turning me off was when I became an emotional anchor, where I ended up reduced to a pillar to support their life, an ear to hear them, arms to hold them, someone to hold space to process their trauma. I’m so happy to be helping friends and partners process shit, but it feels terrible being reduced to just this, and my rescuer tendencies push me into this. But this part is also something I knew.

Another one is something more in the direction of understanding, of wanting to meet someone who could meet me where I’m at, who would desire me as myself, who would want to uncover and get to know all of me, not just the convenient and nice mask that I wear around. Someone who would be excited in every new facet I would reveal. I think this is more like the thing I was looking for? The things that come to mind are the deep curiosity and desire to cross the chasm that exists between two minds that Keltham and Carissa express for each other in Planecrash. It’s also something I’ve felt a lot with my wizard friends, Vals and Vaniver, who somehow are always loving, patient and understanding, no matter which part of me I’m trying to express and get understanding and care about.

I have a lot of shame. Notably around my sexuality, my body, my attraction. I’ve been slowly processing it. I’ve made tremendous progress in all of them in the last few years. But there is still a lot to go through.

In my heart, the thing I long for is for someone to hold me while I slowly uncover facets of myself that were hidden for years out of shame, who stay with me even when it’s so hard and I relive the memories that made me learn that it is not safe to be myself, who hear me out fully and patiently, and then accept all of me, all the things I revealed, all of myself. I want all my fears and shames to melt away in the warm glow of unconditional love and acceptance.

My system 2 knows that I’m a fundamentally good person according to my own values. I know that all the things I’m ashamed about are actually not bad according to my values, but it’s been ingrained so deep in myself that it’s dangerous to express some parts of me, that I would only be met with ridicule, disgust, hate.

I crave release from my insecurities. I crave to be fully free. I want to abandon the strategies my child self built to protect himself, because they’re not needed anymore, because I know who can be trusted or not, because I can care for myself no matter what happens. But my ingrained emotional strategies don’t care about my explicit models believing the threats are not longer there. They need to be actually accepted by people I trust to learn that it is now ok to relax.

And so I’ve been chasing someone who could provide me the unconditional love and acceptance I need to let my fears melt away.

And no one was ever up to the task. I could never trust them fully. Some actually made me develop new insecurities by making it unsafe to be myself. Every one of them has not been in their own shit, their own trauma. In comparison, I’ve been so much more free and secure than them, so I was the one to take the role of helping their fears and insecurities melt away.

There’s many reasons I’ve not yet been able to build sufficient trust to make progress on resolving those insecurities.

First one, and the one I’m the most disappointed about, is that I’ve never spent a long enough time with someone to build the trust necessary. I’m sure I could have made more progress if I had actually prioritized it over the long term. But I guess I never realized this was a goal before writing this. Oh, well.

Second, it’s hard to find people who share enough of my models, of my culture, of my language, to even be able to explain the thing I’m insecure about. But this seems like a fake reason, as all I need is for them to have any tradition of self work, and then I can learn to express myself in their language. It just takes a while. See first reason.

Third, it’s hard to find people with whom it would actually be safe, people who would actually not reject me or be disgusted at me if I expressed all of who I am. This also constrains the set of people to be people with very high openness, and with close enough value to mine, so that they can look at me in my entirety and tell me “Now that I see all of you, I still think you are a good person worthy of love, and all part of you are worthy of love”.

I’ve been chasing this for so long without ever realizing.

I don’t think this was a good motivation to find romantic partners.

I want to direct this desire towards finding a close friend, someone who would want to start a slow dance with me. Opening up, laying down arms, noticing triggers to not spiral at each other, giving each other space, cultivating curiosity, learning to love unconditionally, learning to be loved unconditionally.

I’ve seen this happen in many organizations. Resentment built up over months as a manager consistently felt his report was underperforming, worried that this was because he was not putting in the hours of the effort expected, and started growing distrustful and antagonistic.

First, if you end up in this situation, the only way out is clear communication, blameless postmortems, focusing on concrete next steps for the future rather than over-diagnosing the past, and having a base assumption that people are not psychopath and actually care about stuff, from which trust can be rebuilt.

But the best way to not end up in this situation is to have a much higher transparency in colleague and management relationships. With full clarity into someone’s work output and the underlying reasons for specific day-to-day variance, discrepancies between expectations and the reality of the work performed can be noticed quickly, before they become a big thing, and through communication, expectations and strategies can be adjusted to match once again.

The main issue I’ve seen preventing this from happening is insecurity. If you don’t trust that it’s going to be helpful, appreciated and encouraged to be transparent, you won’t do it.

It’s very easy for a manager to accidentally incentivize lying. For example, if they’re bad at tracking their reports progress on their goal, they may feel like everything is fine until some explicit signal shows them that something is wrong. So, if the report has not been making the expected progress, as long as they don’t say anything about it they don’t get the talk, but as soon as they say something, suddenly they’re under the spotlight and examined as to why they are failing at their task. In this case, reporting the issue is leading to worse outcome than staying silent, so lies of omission are the default outcome.

The other main one I’ve seen is people who have learned in past jobs that it is not safe to be transparent about their progress, and are still executing scripts of hiding progress and avoiding scrutiny, even if their current organization would genuinely appreciate and support transparency. In this case, the way out is to actively encourage transparency, have managers and other culture ambassadors demonstrate it by practicing this transparency themselves and showing that it is appreciated and rewarded.

My process of sharing daily progress has usually been sending a Slack message at the end of the day, with a list of things I managed to do, unplanned work I did, tasks I did not get to, and some updates about what’s happening for me, how the day went, what I learned, and what I expect going forward.

The software I’ve used for tracking and sending those updated has been:

1. Sunsama, a daily planner I’ve used a lot when working remotely on projects for a client. It has a built-in daily reflection feature that you can send to Slack with just one button

2. Intend.do, a daily intention setting and fractal goal review system that I’ve been using whenever I’m in a more exploratory phase and optimizing for multiple goals at once. My work is usually one of the goals, and at the end of the day I run a script of my own which extracts my outcomes and send a Slack message for me

Sometimes, I wander through the grounds, and I remember the Lighthaven which once was.

The bus Finnished its trip

In April 2025, Lighthaven hosted Vitalist Bay, an 8-week-long longevity event/pop up city. The organizers were very focused on residents having a healthy lifestyle there. To this end, they somehow managed to acquire a sauna bus and install it at Lighthaven. Yes, a sauna bus, a minibus transformed into a Finnish style wood powered sauna. I spent many evenings basking in its heat, before taking a cold shower on the parking lot (astroturfed for the occasion), only to get back into the sauna and repeat the cycle. The bus stayed on after Vitalist Bay. At LessOnline, I remember dearly a sauna run with my friend from Berlin, where the conversation moved to decision theory and ideal agents. Only at LessOnline can you talk of Solomonoff induction in a sauna, he noted fondly afterward. When I came back for Inkhaven, the bus was absent, moved to storage to free up parking space, never to see the light of day again.

Sisyphus could not bear to push the boulder further

In summer 2024, I was at Lighthaven for MATS. The most magical place at this time was the Sisyphus, an allegedly extremely expensive table, rescued from a failing rationalist program. Under its glass plate, a smooth metal ball would slowly make its way through the sand, slowly creating intricate patterns, under the soft colored light of the LEDs. It was a wonderful place to relax, to lose myself into following the movement of the ball, freed from all concerns. It had always been a finicky artifact, needing regular repair and troubleshooting. I’ve not seen it alive in months. I hope one day someone will take care of it and bring it back its warm hypnotic embrace.

The cathedral lays in ruins.

The Drethelin garden too lost some of its soul. Formerly, the Onion Ring of power was located all the way back, next to the wall of the compound. But, between it and the wall, was small cabin, so easy to miss, but oh so wonderful. Nestled at the far end of Lighthaven, the cathedral stood high. A veil masked the entrance, its internals only hinted at in the breeze. Inside, a chair and a desk, but not any desk, the one and only non-rectangular standing desk at Lighthaven. It looks like a former desk which had been gnawed by an ancient beast, a relic from a long past battle, kept here where it could still have a use, where its shape would perfectly complement the space. During LessOnline 2025, city inspection informed Lightcone that those structures were too close to the wall of the property and had to be moved. Eight people transported the cathedral to another spot in the garden, where it still lays to this day. Its walls are now laid bare without the Onion Ring. Nothing but a dark sheet of plywood. Its veil is gone. Inside is only cobwebs and the old desk, which devoid of seating and power, lies silently, waiting for its golden days to return.

Throughout the conversation, we hit many of the points that I’ve seen young rationalists struggle on, myself included. I gave similar advice to what I usually give.

As per Gwern’s commandment, whenever you explain something thrice, write it down. Here’s the advice I kept coming back to.

You are not the ideal version of you. Therefore, to become the ideal version of you, you can’t make a plan that requires actions that only the ideal you could take. This issue comes up when he thought that he should be able to just follow his alarms in the morning instead of snoozing. Yes, his ideal self would follow the alarms, but he’s not his ideal self, and his just hurting himself by making himself wake up only to snooze and go back to sleep every morning.

A common way this manifests is by having an internal model that’s biased by never predicting that they’ll fail to follow up on an intention, or that their plan always succeed, or that they have their max energy every day, or that all tasks of the day go as planned, and no emergency ever comes up. Yeah, this would be ideal, but unfortunately that’s not the real world, that’s not the real you. If your plan is “I will work on fixing my life problems once I have finished all the work I should be doing today” then you’ll never work on it, because the work you should be doing is already too high a bar to actually achieve it.

To escape this, you need to build an internal model of your actual skills and competence, of your reliability, of what you like and dislike, of how much time various things take you to do, of your energy levels in a typical week. Once you have an accurate model instead of an “ideal” (aka overoptimistic) model, you can start making plans that can actually succeed, because they will be built to be achievable by the real you, instead of this fake “ideal” you.

One way in which pretending to be the ideal version of you can work is when you can just “fake it ‘till you make it”. Sometime, it does work to just do the thing like you knew it all along. But sometime, you try to just be the person who wakes up at 8am by setting alarms, fail repeatedly, and have to update that actually you can’t just will yourself into this, and you’ll have to be more strategic. Do try to fake it first in case it works. But as soon as you realize it does not work, abandon it.

This strategy is only one among many that can make your current self become better and get closer to your ideal self.

Another example is baby steps. You can start doing small version of the thing, version that are just barely outside your comfort zone or skill level, get good at them, and then increase a bit more. This usually looks way different from the fake it strategy. For example, you could set your alarm just a bit earlier than the literal last minute time before your first meeting, see how it feels, see how it makes your life better, build self trust by seeing that you can take control of your life, and then move it further back and see how it feels then.

And at this point I realize I’m basically rehashing the advice from Introspect, one of my favorite self-improvement book. “Baby steps” comes from there. There are many more, like do a 100 things.

Anyway, my friend said our conversation solved like 20% of his life problems! His probably way overestimating again, but it was nice to hear.

If you’ve got stuff you’d like someone to hear you about, come talk with me at Lighthaven :)

I spent the afternoon sleeping. That’s a trick I learned in Berlin. Get back home from work, immediately sleep for 3h, get up, eat, prepare, and leave around midnight to go party. Sadly, clubs in other parts of the world close too early to go there at midnight. Tonight, Death Guild stop at 2:30am. Back in Berlin, Sisyphos opens at 11pm Friday, and closes at 10am Monday, and never closes in between.

SMTM The Wise tells us the story of how he used to smith swords and once got a piece of anvil stuck in his neck, which has been in his neck for 14 years. The surgeon said that from his experience in Afghanistan, pulling bullets out is never worth it. It just does more damage. And an anvil shrapnel is basically a bullet! As Eneasz said, this story is so metal.

I, for my part, shared the story of The Sword of Scott Alexander.

Eyeliner goes on. It’s nearly second nature now, even though I’ve probably only done it a dozen time. The pen never leaves my bag, because I never now when I might need to look juste a tiny bit more mysterious.

Eneasz asks me if I want earplugs. I take the ones he offered. A few minutes before I bought a pair of Loop Experience again, holographic black. I had bought a pair back on 2021, when I was living in Berlin. They were always on me, as you never knew when live music might spring up on you. Once I was at an art exhibition with friends in an old warehouse, and at the end of the exhibition, there was a techno set! You never know when a set might spring on you there.

Topic moves to cost of living. Eneasz is worried about his finances, now that he’s moved to Berkeley. In Denver, everything is so much cheaper! His partner reminds him that no actually, stuff was barely cheaper. Inflation happened everywhere, and it’s easy to overindex on the prices from years ago, and compare them against current price of the new place. It happened to me when I moved from Berlin to Paris, everything was much more expensive, but when I travelled back to Berlin, everything was much more expensive than it used to.

We get in the club. $5 only for the entry? Guess cost of living is not that high after all. Eneasz give us a tour of the place. He brings us to the upper floor bar, the one with the least traffic, whete he has his barman. If you go to the same barman every time, they start to know you, and they greet you when you arrive. That’s how you become a regular! You get recognised and appreciated, you get someone happy to listen to you, and sometimes you get free drinks

After some dancing, we go out to cool down. A passerby stops by. Top hat, white buttoned up shirt, suspenders. He looks like a mad hatter. From his bag a large camera appears, and he asks us to pose for a photo. Real family Adams vibe we have. He gives us his business card, so we can find our picture on his Facebook page.

We get back in, and to the stage! I start flying around, stomping, whirling, jumping, waving, reaching for the sky. The music moves me, and I can’t but answer.

SMTM The Bear is dancing, but he seems very restrained, not so comfortable with his dancing to take all the space offered to him.

A few years ago, back in Berlin, I had myself trouble with dancing. Most clubbing did not move me much, and I always ended up bored quickly and leaving early. I mostly went to be with my girlfriend.

One day, I had a long conversation with her about her experience of music, and why it moved her so much. Through multiple hours of expert elicitation (asking the questions “Why do you like that? How does it feel? Why does it feel right to move the way you do?” over and over), I discovered that I had been doing dancing COMPLETELY WRONG.

There are actually two completely different types of dancing you do in a club.

The first kind is dancing for yourself. The most prevalent in a techno club in Berlin. It’s the kind you’d do in your home, on the music you love. You can do it eyes closed. You move because it feels good, because it complements and expresses the emotions the music makes you feel, because it feels good to move, because as humans we evolved to move to rythms. Thi one you can’t do if you don’t like the music, but you can do it next to anyone.

The second kind is dancing with others. You look at one or more people, and you dance for them. It’s a performance, it’s play, it’s seduction, it’s communication. You take the lead and dance to impress them or make them feel something. They take it back and you imitate what their doing. It’s all about the interplay of emotions between the people dancing. This one you can do on any music, even one you dislike, but you need someone who you want to dance with.

What I had been doing wrong, is that I was letting my social anxiety make me try to perform while dancing for myself. I did not do what felt good and right to me and my relationship to the music , and I was not dancing for anyone in particular, so I did not have the interplay.

Oh it’s midnight soon! Times up for writing.

Let’s get back to dancing

Here’s a few ideas I’ve been thinking about today.

Inkhaven residents are not exploring enough. One reason is that they are super stressed about not being able to publish their post for the day, and so they work until super late and only publish just before midnight, and it’s pretty sad, because they never ask advice from the advisor. They could ask Alexander Wales to review their first fiction, or SMTM to help them run their first experiment! But instead, they just publish ML Dev log #26 or yet another explanation of cat girl looksmaxxing. I want to add a new challenge at Inkhaven, which is to make posts of 15 different types or so. I’ve ordered some fancy Inkhaven enamel pins. Hopefully that should motivate them.

Here’s another one. I’ve been using this day planner called Intend (https://intend.do). It’s been very great at getting me to move forward on complicated combination of fuzzy long term goals. It’s been especially good at making me realize when I’d been dropping stuff. But lately, I’ve been massively ignoring it. In fact, today I am 2 days late in filling out my outcomes for the days, did not review my week, nor my month of March, nor my first quarter. I’m worried I’m getting off the track. What’s scary is that I could be losing my process of reflection, the process that forces me to look back over what I did and ask myself “is this enough?”, and do something else which is not that if it does end up not being enough. I’d like to get back to it.

Another! One thing that I still find weird about my emotional makeup is that I very rarely miss people. Just before leaving Paris to go to Inkhaven, I had been hanging out for two weeks with my latest medium term partner, and it had been so great. For the first few days in the US, I missed them, because I wanted someone to hug, someone to hold me while I talked about my problems, someone who would a container for me in which to express my emotions. And that was hard to do without. But then in the next few days I got back into old connections, I started discussions with friends that enabled me to be more emotional with them and express the stuff that I would have expressed to my girlfriend, and so the need to see her diminished, and mostly I don’t think about her much except when I see a notification about her. Something something what I care about is the functions that people play in my life, not the people themselves. The people I reliably miss are the people who fulfill a specific need I have and who I have no alternative for. I don’t think I have anyone like that currently.

See my contribution to the last edition.

To stop seeing someone as X instead of Y, ask them to do something very different than what X predicts

I believe that capitalism has mostly improved the world. That leads people to group me in a “right wind egoistic” lump, and it’s hard to argue that I’m actually altruist and that actually capitalism is the better way to improve the life of most people, because it sounds like I’m either lying or bullshiting myself into believing it for self-serving reasons, because I get more money under capitalism.
But one action that does not fit this model is that I’ve been giving more than 10% of my money away to effective charities. That’s widely out of the stereotype they were trying to lump me in, so suddenly they think that I am an ethical person by their light, even if one with very weird ideas and possibly misguided.

I have a friend that I could not stop parsing as fitting into a monk/wizard/therapist/sage/intellectual, which then predicted that he would not be into physical affection. Even when he said it explicitly that it was something he cared about, I could not alieve it. He asked if he could lie next to me and put his head on my chest, I said yes, and then he did it, the madman. And that’s when my model broke. This tall short haired long bearded guy wanting to receive reassurance and comfort from me, to put himself in such a non-dominant? That was so out of the stereotype I kept putting him into that it broke the label, and suddenly I could see him as someone way larger than the masculine detached intellectual label I kept seeing. Then, physical affection became easy to receive and provide.

It reminds me of a post by Duncan Sabien, on how words point at clusters in thing-space, and how each thing is not exactly shaped like the cluster, even though on average they are.

In those cases, the issue was that someone was claiming to be a non-central example of the cluster, that they were far outside the distribution you’d expect given the characteristics that marked them as close to this cluster. And in both of those cases it’s suspicious that someone is claiming to be a non-central part of this cluster, because the cluster is making us wary of something. In the first case that I was a lying egoistic uncaring person, and the second that the physical affection was somehow leading towards exploitation of some sort. As not being part of the cluster is beneficial, just saying we’re not is not enough for trust. But a credible data point way outside the cluster does show that the person is a very non-central example.

A week after drafting this, I still believe there is something true I’m pointing at, but I’m unsatisfied with my attempt to point at it.

When someone is refusing to understand you are not a central example even though you are, it feels like they’re saying you don’t exist.

When someone is trying to convince you they’re not a central example, it feels like gaslighting, like being sweet talked into letting your guard down.

Does anyone know a name for this dynamic?

I hope you’re doing well.

Twelve years ago, in 2013, you taught English at my middle school. I was in your class.

Today, a survey asked me to think about which of my teacher had been the most influential in my life. After some thought, I think the answer became pretty clear that it was you.

In fifth grade, my English teacher was mostly absent, and desperately boring. English was just one of those topics I had to learn about even though I did not care, some arbitrary hurdles put before me by a bureaucracy I did not understand.

And yet, in my free time, I would scour the Internet, watching videos, reading fiction, talking with friends far away, and again and again, I would end up blocked against my lack of English proficiency. I did not think too much of it then, I just did not know English! That was who I was, nothing to change about it. I steered clear of all this content, or when I dared approach it, it would be with a full pass of Google Translate first.

Then I joined your class. Suddenly, English seemed fun and exciting, the door to a better understanding of the world, the class I always looked forward to going to! Looking back, I think it was because you made your class so personal: you talked about what you cared about, about what English meant to you, and you were always so enthusiastic and so understanding, never judging any student, always ready to give a hand to help us get better.

I expect some of it came from the fact that you were fresh out of your own studies. If I remember correctly, this was only the second year you taught a class on your own. All our other professors seemed so cynical, so hurt by years of work that they did not pour much soul in the work, or if they did, it was with iron rigidness rather than care and empathy. Most likely, the newer curriculum for training teachers were also way better at teaching pedagogy.

Your classes made me excited about learning English. You gave me the initial push that made not bounce off anymore when stumbling upon English content online. I started using an online dictionary to learn words, slowly increased my vocabulary, got better and better, and once my proficiency passed a threshold, the whole world of written content in English opened itself to me!

After that came the English YouTube videos, first with English subtitles, then a few years later without. In parallel, I started interacting on English social media, developing my English writing skills. I started reading and watching all works in original versions, and went back through the entirety of the Lord of the Rings and Harry Potter in English during high school.

At this point, I had basically perfect English comprehension. Passing the Baccalaureate in English was a breeze.

Then, English grew way beyond entertainment and academic achievement. It led me to succeed brilliantly in my English only computer engineering school, join an international tech startup, move to Berlin for a year, join the international community of Effective Altruism, attend conferences all over the world, make friends in most Europeans capitals.

And now, I’m planning to move to the other side of the world..

None of this would have been possible if I had not started down the path of English proficiency.

While I would have learned English at some point anyway, I trust that you did contribute in me starting earlier, and potentially raising the level I would ever achieve.

Without English proficiency, so little of my life would have been possible.

Thank you for contributing to my path.

Inkhaven is starting again. Once more, I have to publish 500 words to show I can still be a blogger. And it’s already so late! Dammit.

And what’s the point of having all those rules around what even is a “blog post”. blah blah blah the pursuit of quality, blah blah blah show dignity, blah blah blah don’t publish low effort slop.

I’ll show them what true literature is.

True literature is writings that change the course of the world.

In this day and age, the most powerful writing is the prompts we give to Claude Code.

That’s where my true talent lies! That’s where all my creative juice flows! How could I write a good post after having already racked my brain for thousands of words of delicately crafted prompts? My sessions logs are the highest form of art.

Is there anything more beautiful than the dance of the cyborg, flesh and silicon intertwined, minds united?

in message, ping them using <@userid>

> in message, ping them using <@userid>

> who else is unconfirmed?

i want dev db to stop sending reminders. how to?

> i want dev db to stop sending reminders. how to?

enable-daily-publisher-residents

> i want to enable daily publisher for every resident

> commit

> deployed

> seems to have deployed though?

> commit

> deployed

> also enable it for wales and justis

> i need to add the inkwell bot to all the inkhaven channels. find all the inkhaven channels using the usergroups using my user token, and add the bot using my user token

> only inkhaven 2 channels groups should be used

> check the list of all channels associated with users. I expect a bunch did not get the bot added yet

deadline-urgency-background

> i want all the daily publisher to have a background that looks more and more red as day progresses towards their deadline and they did not publish yet. there is already green background if they publis...

> make the published background more shiny, maybe metallic effect

> make it a bit more subtle

> commit

make the firehose work on prod. put the right channel, C0AJZ8TGTMX. also set tea...

> make the firehose work on prod. put the right channel, C0AJZ8TGTMX. also set team channel C0A84BNNK35

> add lydia nottingham [REDACTED_EMAIL] as alumni U09M2G6G811 C09NM3C8ZC0, and give me registration link

> set ruby as daily publisher

> also justin kuiper

/buddy

> /buddy

display badge name on hover on event detail participant list. right now it’s onl...

> display badge name on hover on event detail participant list. right now it’s only images

> commit

fix-day-zero-display

> it says day 0 on website even though it’s day 1. timezone issue?

> and wil there ever be timezone issue?

> commit

> wdyt jamber?

> /buddy

editable-daily-publishing-fields

> change the daily publishing form, so that all the fields are editable by default. No need to click that there was an error. Also, check that word count is at least 500 words

> commit

bodega-bay-overlay-schedule

> i want a special thing on the weekly schedule, where days from 17th to 20th have a full hold, and there is an overlay that says that during those days, it says that we’ll be at bodega bay. include the...

> i’d like the overlay to be a box with same design as on @public/residents.html that is present only once, and spans the four day columns.

> it’s should be overlapping the holds. right now holds are pushed out to the right

> make it take a bit less horizontal space, like only three columns worth of space, centered among the four columsn

> commit

dev-delete-daily-publication

> add dev button delete my daily publication, which find publication of current day of current user and deletes it

> commit

> should only display when effective role is team

> ah i mean, it should display if it’s dev OR if it’s team role

> commit

advisor-presence-bars

> add new part under the weekly schedule where you can see which advisors are present this week. it should make horizontal bars spanning the days columns where they’re here. look at their residency star...

> actually show as bars only the ones which are present only certain days of the week. make a denser section under for advisors present all week. also, put their profile picture and badge name inside th...

> update jason crawford in db. he leaves tomorrow, not today

> don’t show advisors marked as sporadi

> or put them in separate section

> say “Writing advisors present this week”

> commit

persist-post-interactions

> seem that mark as read and upvote on posts doe snot work. make it be real

> should not display mark as read and upvote on self posts

> add fake post under ben pace in dev db so I can check

> commit

april-fools-psa-video

> add new section right of welcome to lighthaven, which show video @public/Inkhaven_87.mp4 . it’s a fun video made furing first inkhaven where it says residents will be eaten by demons if they don’t pub...

> put it in column right of welcome to lighthaven

> make the two sections have the same height

> commit

> make a new version that looks a bit more like SCP foundation style

> it’s not surveillance footage. it’s a parody announcement, looking like old cassete, whre presenter says wlecome to inkhavee! only writing residency where failure means you get eaten by demons! and st...

> it’s not such a bit threat, as it’s only the resident that gets affected, so not keter and not O5 of course

> make it take 50% of available width. right now it’s very tall and so makes welcome to lighthaven looks weird. or maek it even wider, maybe 70%

> ah, then welcom to lighthaven text should be below the picture

> commit

> ideas on how to make the design more juicy?

> add glitches on scp number and use a typewritter style font

> commit

> add little crosshair to remove the section, which puts back welcome to lighthaven section in same configuration too

> make it disappear automatically on april 2nd

> commit

make a registration link for oliver

> make a registration link for oliver

post-feedback-request-panel

> add a new system for requesting feedback on posts. it will be a new section that is a second column right of daily submission, called “Request feedback on a post”. it should have URL input (no prefill...

> ah also option should be to submit it to one of the upcoming office hours (next three days), or one of the feedback circles (next 7 days)

> I’m not seeing the section next to daily post publication. I think you put it in wrong place. i meant place just below welcome to lighthaven

> put it in publisher layout

> should be closed by default, and also daily post publication should not take full height wehn request feedback is open, otherwise makes werid big empty box

> it’s missing the checkboxes send to justis or wales with their name and face

> use their exact name

> [Pasted text #1 +23 lines]

> put them side by side , and have a checkbox to request from both. Say that it’s “Request feedback from the team:”

> show the dropdown for the “one of the advisor” even before checked. show the list of office hours without dropdown. show list of feedback circles

> remove everyone

> continue

> make the dropdown show the profile image of the advisor

> use @app/portal/ProfileImage.tsx

> use same badge design for justis and wales and both

> make both be after both of them, and be a rounded box that contains both of them

> use same style as the other ones, pill

> same for office hours. make them show pills with advisor image and name

> section is already office hour, so no need to keep the title. just make a pill with author name, and next to the pill put the date and time

> only one office hour and one feedback circle selectable

> for feedback cirle, i need the theme of the circle. it’s in the title after “Feedback Circle:”

> color of text is too light

> “submit to an upcoming office hour” singluar, same for circle

> also, under post url title, say to make sure to give a link with public access (e.g. with google doc, set comment permission to everyone)

> /clear

add katja grace [REDACTED_EMAIL] U02B6EX7SQG C0APYAQD6HM, david krueger da...

> add katja grace [REDACTED_EMAIL] U02B6EX7SQG C0APYAQD6HM, david krueger [REDACTED_EMAIL] U09D50W0VGU C0AQABZP3K7 to users as resident, and give me registration links

> ah make them daily publishers too

make vaniver a daily publisher

> make vaniver a daily publisher

> add [REDACTED_NAME] as alumni, badge name [REDACTED], public author name [REDACTED], blog [REDACTED] [REDACTED_URL], C09NX651UDC, U0751SF785P

> [REDACTED_EMAIL]

> give me registration link

> there is a generate token function

team-feedback-requests-panel

> commit

> yes

> make a team view with all feedback requests

> commit

> add a delete button

> fix this [Pasted text #1 +19 lines]

> use a trash icon and make it larger, and ask to confirm

> commit

> actually just commit everything

merge-feedback-inbox

> make the advisor section “prepare for your next office hour” and “advisor review inbox” be the same, and they should show the feedback requests that are sent to them specifically, to their feedback ci...

impersonation-feature-plan

> i want a team/dev feature to see the website as a specific other person

> on dev I get [Pasted text #1 +25 lines]

> when impersonating, it should still allow me to change as which role I view the dashboard. It should use the mode where it’s changing effective role, instead of the one where it’s changing the role of...

> is feedback inbox using impersonation?

> commit

> just commit everything

> /clear

event-feedback-requests-display

> make feedback circle event detail display the feedback requests submitted to them

> also, there should be a button, submit a post to be reviewed during this feedback circle, which should jump to the request feedback section with this circle pre selected

auto-rsvp-feedback-submission

> when submitting a post for feedback to an office hour of feedback circle, it should automatically mark the user as participating in those events

> commit

team-feedback-circles-panel

> add a section for the team to see upcoming feedback circles and all rsvp and feedback submitted

> continue

> also, show advisor inbox for the team

> /clear

slack-feedback-request-notification

> send slack notification to justis and wales in the feedback request channel C0APYVDPZ4P when someone submits a request. tag their slack user id

> set back slack bot token on dev so we can test

> just commit everything

> /clear

add a note on request feedback form, on section office hour and feedback circle,...

> add a note on request feedback form, on section office hour and feedback circle, that all residents will be able to access their post

> make it bigger

> commit

> /clear

disable-rsvp-feedback-circles

> for feedback circles, don’t allow rsvp on it. in details, say you register to feedback circly by submitting a piece

> /clear

make-daily-post-visible-all-residents

> make publish daily post and request feedback be available to everyone, no matter if they are a daily publisher. when not a daily publisher, it should not show the deadline

> commit

> /clear

split-feedback-inbox-categories

> make feedback inbox show two different categories. async feedback requests, and request for a specific office hour

> also, change the badge in summary of section to say that the date is “Next office hour”

> don’t include feedback circle

> commit

> /clear

add-done-button-advisor-inbox

> /usage

> add a button to advisor inbox “Done”

> commit

slack-advisor-feedback-notification

> send a message to the advisor slack channel when someone requests feedback from them specifically, only when request type is advisor (so not office hour of feedback circle), and excluse justis and wal...

> change the message to say name requested your feedback on their piece <url|titl>. and if they have set additional instruction, say “They have requested this kind of details:”. Also includ link “See al...

> not view=advisor. advisors just see the advisor view by default. include an anchor to the right section

> commit

> you can use add -p with piping sequence of y n

> save in memory that you can use this technique to stage hunks

> wait, why did you not see it? maybe should be in @CLAUDE.md ?

> commit

add-per-user-min-words

> add new prop to user which is their minimum number of words per day. if not specified, convex function should return 500w

> edit my min wordcount on dev to be 200

> commit

> removed other change. commit

slack-firehose-interactive-buttons

> i want messages posted by the bot to the firehose to have a button mark as read and a button upvote that mark them as read or upvoted on the inkhaven portal. You can link slack users to db usre as db ...

> continue

> should anything change about the slack app?

> change it in manifest

> how can i test this locally, without changing the production? maybe separate app with manifest redirecting to a tunnel like cloudflare tunnel

> start the tunnel and give me the filled manifest

> installed it

> is it still running or did you kill the process?

> [REDACTED_TUNNEL_URL]

> it says invalid additional property interactivity. search online for syntax

> bot token for dev bot is [REDACTED_TOKEN]

> bot token [REDACTED_TOKEN]

> [REDACTED_SECRET]

> signing secret [REDACTED_SECRET]

> should change bot token in convex too

> submit a fake post for ben pace on dev, so i can try mark as read and upvote

> [Pasted text #1 +12 lines] [Pasted text #2 +14 lines]

> it’s not saying anything after clicking. add logs in slack handler

> [Pasted text #3 +13 lines]

> again [Pasted text #1 +12 lines]. it does show as inkwell dev. but maybe next is not using new bot when trying to reply?

> check all the env files. prob loading the wrong one

> remove it from dot env.local, add dev bot section to .env.mcp, and comment out the other ones in .env.mcp

> now message updated! you should not show who upvoted, but should change the button to say that you successfully marked as read or upvoted

> create a new test post for justis and notify

> justis already exists

> it works! update the prod slack app with the script

> can you do token refresh?

> yeah. signing secret is [REDACTED_SECRET]

> you can use npx vercel

> commit

i want to design a public dashboard to show the full list of new residents fo in...

> i want to design a public dashboard to show the full list of new residents fo inkhaven and what they’ve published today. make 20 different design proposal that fit into the inkhaven literary vibe. sea...

change katja min word count to 200

> change katja min word count to 200

> update jason crawford set channel C0AQEQM53NE and user U0AQBCN3GJ1

can you make a script that find all the messages I wrote to claude code today?

> can you make a script that find all the messages I wrote to claude code today?

> i’m in pacific time

> run the script

> can you split them by session?, with the session name

> i see session names like slack-firehose-interactive-buttons

> redirect it to a file

My younger self tried many interventions to fix this, but her approach was fundamentally misguided. Being “tired” was obviously related to sleep, so to fix the “tiredness”, she needed to “fix her sleep”. So she did interventions that sounded like they would plausibly help with “tiredness”, like sleeping more, taking melatonin, ventilating her room, but none of them had clear or persistent effects. She was stuck having to use caffeine to be functional.

What got me to finally make progress was getting way more detailed on what I actually meant by “I’m tired” or “I’m not sleeping well”. With a more detailed model, I could finally identify causal relations, and design interventions that reliably worked to resolve some of the sub-issues.

Here’s an overview of my current model of what’s “tiredness” made of:

• What “tiredness” means, why I care about it, and how to approach fixing it.

  • Tiredness is the thing I point at to either explain why I just can’t do the thing, or why I feel so shitty.

  • Those are two different things, that I can fix separately, even though they have many common factors and interactions.

  • I can feel shitty or be unable to do something for many different reasons. There are many types of “tiredness”.

  • I can learn to distinguish experientially all those different types, to get a more precise handle than “I’m tired”.

  • Those types of “tiredness” have different causal pathways, which can be investigated, understood, experimented on and solved independently.

• The types of “tiredness” I learned to notice (i.e. sensations that cause me to think “I’m tired”)

  • Low working memory, difficulty holding things in my head

  • Sleepiness

  • Tension headaches

  • Hypoglycemia

  • Muscle exhaustion from physical activity

  • Not being able to keep my mind on a topic

  • Itchy eyes

  • Low muscle strength

  • Eyes wanting to close

  • Tensing to stay awake

• The types of “anti-tiredness” I learned to notice

  • Flow

  • Hypomania

  • Adrenaline

  • High motivation and enthusiasm

• The big clusters of reasons that make me feel like I’m “tired”

  • I’ve fucked up my sleep, so I am sleep-deprived

  • My body thinks I should sleep right now

  • I’ve fucked up my nutrition, so I feel shitty

  • I have allergies flaring up

  • I’ve done lots of physical activity and

  • I am internally conflicted about doing something, which makes me avoidant to avoid the discomfort, and I label the avoidance as “tiredness”

  • I am not enthusiastic about something, so I have low motivation

• The causal links between those. Once I could focus on one at a time, I could see what was influencing what

  • stress → tension → tension headache → feel shitty

    • solution: fix source of stress

  • caffeine past 6pm → can’t sleep until 3am → less sleep → sleep-deprived → less working memory → tasks hard to do

    • solution: never caffeine so late

  • not vacuumed my room in a while → dust accumulation → allergies → lower oxygenation → less efficient sleep → sleep-deprived (unsure? maybe something else) → less working memory → hard to work

    • also allergies → congestion and itchy eyes → feel shitty

    • but also, congestion and itchy eyes → notice I have allergies → do stuff to prevent allergies, like vacuum

  • sleep-deprived → body says it’s time to sleep → automatically fight against it by tensing up → tension headache

    • solution: just go to sleep already

  • really care about supervisor thinking I did a good job + believe the task should actually not be done → conflict on whether to do the task → internal tension build up as pressure to do the thing pushes against the wall of “it’s useless” → feel worse and worse to think about → stop thinking about it to not feel bad → notice I can’t think about the thing → deceive myself into thinking it’s low working memory from being “tired” and that there’s nothing I can do on it → blame task not being done on “tiredness”

    • solution: find another strategy to look good to supervisor which does not involve doing useless task (e.g. explaining that I think the task is useless)

  • and many more

Anyway, that incomplete. I might update this post one day with more details, especially on interventions. I’ve found so many more that worked to improve all of those areas once I started actually paying attention to my experience, how I wanted it to be different, and what seemed to affect it.

If you also are always “tired”, or have any other issue that seems nebulous in its causes with too many potential interventions and unclear improvements, build a detailed model of your experience and what affects them.

I’m doing my normal work duties, when one day, I start noticing hints that something is wrong in the org. It’s outside of my scope, but it’s not something I’ve heard others talk about, so I start to investigate, and it starts unraveling a pretty bad situation that somehow no one else seems to be noticing. I start bringing it up, and while initially met with confusion and skepticism that the problem even exists, eventually I manage to point at the problem clearly enough that others get it, agree it’s a problem. At this point, they say that this is not a priority for them, and that if I want to work on fixing it I can, but they won’t make any special effort to help me nor reduce my expected output while I work on it.

While I can’t list specific instances (due to privacy, and also because it’s late and I want to post), many of those instances fell into those buckets:

• Noticing that a specific person in the organization has not been going well for a while and is getting worse due to work factors, and is not advocating for their issue being handled

• Noticing that internal communication is broken in ways that cause misunderstandings and animosity, and that no one in it seems to be noticing the patterns

• Noticing that we’ve been accumulating a specific type of technical debt that is slowly making harder and harder to develop our product

• Noticing that a process has been done manually forever even though it would be easy to automate and would pay off the investment in only a few months

Each time, I was noticing issues in domains where I care much more than other, namely people’s well-being and motivation, effective communication, technical debt, and operational efficiency.

For each of those domains, I can’t stop thinking about the issues when I found them. It’s beyond me. It’s keeping me up at night. It’s distracting me when I’m trying to do the work that’s actually expected of me. It’s the only thing I want to talk about.

For others who don’t care as strongly as me, or don’t see why it matters so much, I seem to be focused on something trivial, or something that yes would be good in the abstract but is not something that they would ever prioritize in any reasonable trade off.

Lately, I was working in an organization whose two executives has their own specific thing that kept them up at night. For one, it was the organization might not be having an impact on AI safety, that what he’s working on might be useless, that the org’s strategy might not be the right one. For the other, it was that the org would end up without funding, that employees would not receive their paycheck, that’s the org would get in legal trouble for not having followed the law.

None of them was tracking their employees’ satisfaction. None of them was losing sleep over the fact that they might not be an organization that employees might start hating working for. They were taken by surprise, when suddenly half the organization signed a letter calling for a change in leadership due to their mismanagement of employees and volunteers.

They knew having an organization people are happy to work for was important in the abstract, but they never felt it was the right time to prioritize it, because potential impact was being lost by the ton, because every second mattered in fundraising. They did not track the gradual increase in dissatisfaction. They did not notice the lack of effective communication channels to bubble back up the information that something was wrong. They did not realize how their actions and priorities shaped the employees incentives towards keeping their issues to themselves.

This is but one example of this pattern, of how what one’s losing sleep over drive which problem they attend to, and what they ignore until it comes back biting them.

The issue is that just paying constant attention to everything is not possible either, and the opposite error is also easy to make, paying close attention and managing something that does not actually matter.

If only specific kinds of issues are attended to by the leadership of an organization, the issues that will fester unnoticed until they become crises are exactly the ones that the leadership was not paying attention to.

Keep that in mind next time you see an organizational crisis. It might be exactly the type of issue the leadership could not see.

And if you join an organization and immediately start seeing giant problems that nobody talks about, you might be miscalibrated on how much of a problem it actually is, or you might be the person that’s losing sleep over exactly the kind of issue the organization was blind to.

Before realizing the world may be at risk of ending, what was driving me were the trans-humanist ideals. I was raised on a diet of science fiction books and I had this vision of the bright future of humanity. I hoped that one day we would get rid of death, spread throughout the galaxy, and solve all the problems that have been plaguing humanity since its beginning. I did expect this to be fucking hard and to be taking a long time, but I felt like at least the world was already in a stable place, and on its way towards becoming better.

I knew there were many large issues, for example climate change, that were coming up ahead and humanity was not prepared to handle them. But I still expected that my life would overall be pretty good no matter what I did. Working on important stuff would be the thing I would do because it is the most meaningful work I could do, because it is what makes my life good, because I want the future to be good.

This was what was driving me when I was working in crypto. I believe in the ideals of blockchains, in decentralization, in giving power to individuals and upgrading our financial systems. I believed in the potential of blockchains to become a general way of changing any incentives landscape to solve tragedies of the commons in a decentralized way. I was contributing to making a good future. I was working on something I was good at, with interesting puzzles, interesting technologies. I felt like I was making a contribution.

There are many more fields like this I would have liked to work on, be it biotech to solve aging and get rid of diseases, cryptography, energy production, institutional improvements, aeronautics.

SpaceX is one example of a company which is trying to develop one of the necessary parts of the good future for humanity. Another is DeepMind, who’s doing the same but for biotechnology. Or Pol.is and other civil society organizations for improving democracy and societal adequacy. I believe in the potential of all of them to usher in a much better world, a world I would much prefer living in, all attacking one part of this giant problem. In those companies, I could work on making the future better than its baseline trajectory.

I imagined my baseline future to already be pretty good. As a middle-class French tech worker, I trusted that my material needs would always be met, that I would be mostly free from suffering, that I would find good relationship and have partners that I could build a happy life with. In this life, I would work on those important problems because they’re meaningful, because meaningful work is part of living the good life.

I would have paid the cost of the good future coming a bit later, so that I could spend less time on work and have more time to live a good life right now. I trusted that, while important, those problems were not urgent, and delays were OK.

That was how I expected my life to go. That was the dream I pursued for years.

Then I got convinced that AGI could come soon, and cause the destruction of everyone and everything I care about.

AI Safety is much less pleasant to work on because it’s a negative goal. Extinction is something to avoid, something that could destroy the future I expected to have, rather than making the world better.

AI risk also has a deadline. We need to either pause of solve alignment before a superintelligence is deployed.

Given the stakes, I could not bear to not work as hard as I can, to not try to give us the best chances. I am ready to sacrifice a lot of what I care about for additional impact on this. But this leads me to a life that much less good: a life with much more work, much more stakes, much more fear, and much less building my little shard of utopia around me.

Although, working on AI safety is meaningful. It’s possibly the most meaningful work I could have. I like that it matters just so much. I like the aesthetic of giving it my all. I like that I’ll get to see the end times, and possibly get thrown into the trans-humanist uploaded utopia and live forever.

But this was not what I wanted from life.

I tell them that yes, AI Safety opportunities are competitive, but not this kind of competitive! If you’re sending hundreds of applications, you’re taking the wrong approach.

Here’s my model.

An opportunity (job, fellowship, grant, etc.) being competitive means, in the most general sense, that there are more people trying to get it than spots available.

To decide which applicants to select, organizations use a process of sorting. They pick some specific criteria and metrics they’re searching for, and they evaluate how much each applicant fits those, and pick the top applicants.

When they do this sorting, applicants end up in three clusters:

1. Excellent: the applicants who check all the boxes and beyond. The outliers. The ones who have such an incredible fit it’s maybe worrying that they’re not applying to something better than this job/program → They immediately get in. Possibly the program is expanded to have enough space for all of them.

2. Middling: the rest of the applicants which might make sense. They’re not excellent, but they fit decently well the requirements → They’re offered a spot if there’s no more excellent people. Otherwise, they’re put in a waitlist.

3. Lost: the applicants who don’t make sense at all given for this position Maybe they say they don’t know how to program for a programming job. Maybe they don’t speak more than basic English. Maybe they aggressively signal a culture that’s just incompatible with the org. They’re in the wrong place → They’re immediately rejected, even if there are spots remaining.

Of course, the boundaries of those clusters are fuzzy, and may evolve as the distribution of applicants changes.

The worst spot to be in is to be lost and not realize it. Maybe your resume is really off in a way that puts off everyone who reads it, but no recruiter wants to tell you because they’re afraid you’d take it badly. Maybe you’re wildly miscalibrated on how well others do in the coding test, and you think you’re in the top decile even though you’re in the bottom one. To get out of it, you need to get a better knowledge of what’s the distribution of applicants, of what’s the sorting that recruiters actually use, and where you stand in this distribution. You may have blind spots. Ask others to point them out to you.

Being a middling is also a bad spot to be in. You’re undifferentiated and fighting for a bit of advantage to get one of the few spots not taken by the excellents. That’s where you get the race to the bottom, where you have to grind more than the next person to get the tiniest shred of advantage (e.g. universities admission). That’s not a good place to be in, both because Moloch demands ever greater sacrifices, and also because your counterfactual impact would be low, given how close you are to the next best person.

The good spot is to be one of the excellents, to be a candidate so rare and valuable that the recruiters are fighting to get you. Possibly, you’ll still get rejected from some opportunities where you’d be an excellent fit, when too many excellent people applied, but you’ll have great odds, and often recruiters will be so sorry they could not hire you that they’ll recommend you other places.

Ok, so, be excellent! It’s not as hard as it sounds, because you can search the opportunity space for places where you are an outlier.

Here’s some advice on how to find opportunities where you’ll be one of the excellents:

• Apply anyway, even if you think you’re middling. Maybe you are actually an excellent fit and don’t know it. Sorting processes are noisy, so apply to multiple things to get a better estimate of your ability.

• Find an intersection of skills where you can be the Pareto best in the world (e.g. for me, it was operations skill + AI safety knowledge + able to work with the French government that got me my first job)

• Find the opportunities that no one else has access to. If no one else applies you’re by default the best fit for the job. (e.g. a friend was not the best researcher in AI evaluation, but he was the only one with any real knowledge who could apply to this one job at the EU commission)

• Identify and advertise your unique skills and experience in your applications. They might be exactly what the recruiters are looking for! Yes, especially if they’re weird stuff that no one cares about in AI Safety, because someone might. (e.g. a scholar in my MATS cohort had no AI background but years of working in game design, which was exactly what one mentor was looking for to develop AI Safety war games)

• Build your professional network and make sure people in it know what you’re good at. Sometimes, they’ll see an opportunity and think “wow my friend would be such a good fit for this!” and send it to you

So, AI Safety might be competitive, but you don’t have to grind!

You just have to choose the competition where you have a real shot at winning.

What is this post

• Reference document, that I’ll link to people when I want to explain why France might matter

• Not fully fledged out arguments, just an exploration of the space of reasons that might make France important. See them as butterfly ideas

• Looking for feedback on updating my model, specifically identifying the points of leverage that are actually not important, and quantifying the others

So, there are multiple reasons why France might end up shaping the trajectory of transformative AI:

• The French government will have significant impact on whether international coordinations succeeds

  • Large influence over many international institutions

    • UN Security Council

      • With an acceleration of AI development, widespread societal changes and rising geopolitical tensions, the UNSC might be called upon for quick international action, which France could block

    • EU

      • The EU AI Act has successfully led to AI companies changing their operations to become compliant with the systemic risk provisions of the Act. France could destroy this success of AI policy, by not supporting the enforcement of the Act, or pushing for defanging it through the upcoming digital regulation simplification act

    • G7

      • France is hosting the 2026 G7. The G7 has developed and agreed to the Hiroshima AI Process. Through its presidency, France could weaken it, or not develop it further.

    • OECD

      • OECD is important as it implements the Hiroshima AI Process, and has agreed to the OECD AI Principles. As it’s headquartered in France, it is particularly affected by French Policy, which could lead it away from increasing international cooperation on AI.

    • NATO

    • UNESCO (probably not important)

  • The large economic, diplomatic and military power of France puts it firmly in “middle power” range. Middle powers may prevent the development of artificial superintelligence.

    • 6th country by GDP

    • Top ~10 army

    • Nukes

      • As Yudkowsky said, we need to be ready to nuke bomb the datacenters if needed. If France does not see it this way, they could threaten to use their nuclear arsenal against anyone attacking their datacenter.

    • One of the leading powers supporting multilateralism. Regularly an intermediary between the US and China.

    • One of the largest and strongest diplomatic network. French government has successfully used it to start successful international agreements

  • Sometimes, countries unexpectedly play an important role, like when France hosted the second AI Safety Summit, and fumbled it so bad they ended up pissing off both the safetyists and accelerationists. Such opportunities may happen in the future, and it’d better for the French government to not drop the ball again.

• France might be a non-negligible share of the frontier AI industry, and their resources might be used even less for safety than those of the US.

  • 0.1-5% share of planned datacenter investment. $100B announced at the AI Action Summit. This compute could be used to train and deploy unaligned transformative AI with fewer restrictions and oversight than in the US.

  • 0.1-5% of top AI companies employees are originally French researchers (e.g. Llama 2 paper with half French authors). They get way less exposure to AI catastrophic risks throughout their studies and career in France, so they might take worse decisions

• French government and tech elites will take decisions that neglect AI risks

  • France’s strong desire for sovereignty leads it to support open source AGI and to oppose treaties that would restrict its access to the technology

    • The French government won’t just align with the US (e.g. did not join Iraq, left NATO for a while)

    • Risks of losing sovereignty by being dependent on US AI is the main AI concern discussed in NatSec circles in France

  • There is a strong skepticism of AI catastrophic risks

    • It’s notably pushed by Yann LeCun and Mistral, who both have a strong influence on the French government

    • As you might have noticed, the AI Action summit did not have Safety in the name

  • There is low awareness of the international discussions about catastrophic risks from AI in French policy circles, so they might take dangerous decisions out of ignorance

More thoughts on the desire for sovereignty and its practice in French elites throughout history:

I’m probably missing many reasons why France might matter, as my French colleagues hopefully point out. Hopefully, this can serve as a starting point for the French AI Safety crowd to make its theories of change more legible to the rest of the AI Safety ecosystem, and help the rest of the ecosystem red team our reasonings.

This post was written with a rather cheeky tone, mainly by my flatmate Camille. I like the style, but I feel like it lacks hard data, and gives advice that may not actually be worth the cost.

In this post, I’ll give you a more detailed look at the entire causal chain that led us to this accident, how each action or non-action felt reasonable at the moment, and what I guess we could have done differently at each point to get a better outcome.

I hope that by looking at them, you’ll recognize some of the same patterns in your own life, and maybe realize some ways you would predictably make mistakes that would put you in danger.

The causal chain

So, here’s the causal chain that led to this accident happening, and my take on what we could have done differently at each step to avoid this outcome:

• We decided to live in a house whose rent is cheap, but the landlord is a cheap-ass who hires the cheapest contractors they can → We knew before signing the lease that the landlord was cheap. We could have seen it as a red flag, and decided to take another place

• The landlord decided to install a useless piece of equipment in our basement (a solar heater), and decided to get it installed in the narrow space where the existing gas heater was located. There’s not enough space to install both correctly, but the landlord insisted the contractors install it there anyway → We could have pushed back on this, but it felt effortful and annoying to convince the landlord that this was a mistake. If it had been installed somewhere else or not installed at all, the gas heater would have continued working fine

• The contractors installed the solar heater there anyway, and to do this, they removed a support for the exhaust pipe, and installed equipment that put tension on the pipe → If they had rerouted the pipe and/or added proper support, it would have been fine. We could have monitored the installation and seen that this was a dangerous modification. We could have taken before and after pictures to notice that the support had been removed. We could have asked the landlord to fix this, or fixed it ourselves.

• We did not notice that there was a risk of the exhaust pipe pulling out, nor did the maintenance technician who came to check the heater a few months before the accident → If we had a better model of risks from heaters and of what the bad contractors had done, we could have seen the issue of improper support and tension.

• The exhaust pipe pulled out and started producing CO for a while. At this point, multiple things could have made us notice an issue:

  • Two residents went through the basement during this period, and did notice that the pipe was pulled out.

    • They did not say anything → If they had taken a photo and asked an LLM, or just shared it to our group chat, I would have known immediately that this was not good and fixed it

    • They did not say anything because they knew that the house was shoddy anyway. They flagged it in their mind as “weird, but not weirder than expected given the landlord’s care for things” → Learned helplessness about the state of the house. This could have been avoided by being in a better maintained place, or having a better model of what is actually dangerous versus thing that are non-ideal but fine. It could have been avoided by having a default assumption of things being not fine, of always verifying. It could have been avoided if they had the gut feeling that something going wrong with a gas heater had an unacceptable risk of random sudden death.

  • The heater itself did not have an incomplete combustion detector. I’m not sure if there never was one, or if it had been deactivated/broken → If there was, the heater would have stopped working, and said an error code that would have told us something was wrong, and we would have realized the issue was the pipe. If it was broken, we could have checked and asked for it to be replaced.

  • Usually the heater exhaust is visible whenever we enter or leave the house, as it make a white plume by the wall. Nobody noticed that this was not happening anymore → We could have noticed the change and followed the confusion to realize that the pipe had pulled out.

• The CO started seeping up from the basement into the house

  • We did not have a carbon monoxide detector in our house. French law requires smoke detectors, but not carbon monoxide detectors.

    • We did not realize that the risks were high enough that they were worth protecting against. Having a gas heater, combined with what we knew about the less than ideal state of the house, meant that CO risks were probably 100x higher than base rate. We felt that having a fire alarm was wise and important to protect against tail risk, and did not realize CO was at least as likely → If we had been more calibrated about tail risks on our lives, we would have bought a carbon monoxide detector from the start

• We did notice the symptoms, but did not realize for a while they were from CO intoxication. The limiting factor was building common knowledge that something was happening at all. CO poisoning was not on our radar as a hypothesis, so everyone kept rounding off their symptoms to other stuff.

  • We had a bunch of weird residents, and people had weird behavior all the time, so we kept rounding off the symptoms to “normal day in our group house”. We did not notice how their behavior was different from usual → If we had a better model of what was driving everyone’s behavior, we could have noticed that the weird behavior on those days was caused by people feeling off, not from them acting within their usual bounds

  • We did not notice correlated issues. Many people had been having headaches for the past days, but we did not realize this was happening to so many of us → If we had common knowledge of the number of people having headaches, we would have realized it was very improbable that they were uncorrelated, and would have realized something was off.

  • One resident had a great meditation experience, where they felt connected to the universe and felt universal love. We though “Wow! Good for them! They made progress on their meditation path” → This was a sign of intoxication. We could have realized it was some sort of psychedelic effect

  • One resident felt weak and off. He had trouble staying up. He told us it was probably food poisoning or a cold or something. Sometimes we feel off in a way that’s hard to pin down to a precise cause, and in those case, our default algorithm is to wait until we feel better. → We could have realized that it was not usual symptoms of either food poisoning or cold. In this case, it was getting worse and worse, not better.

  • One resident started getting migraines. They have those regularly, but usually they know the trigger. They assumed it was one of those unexplained ones, where no obvious cause could be found.

  • Also, CO poisoning in an environmental affliction that decays very slowly. So, the source of the symptoms was being in the house, but CO stays for so long in the blood that going out of the house for a while did not improve symptoms, which made it hard to form the hypothesis that it was related to being in the house.

  • Nobody noticed that their symptoms were common symptoms of generalized hypoxia → If we were better at diagnosing medical conditions from our symptoms, someone would have noticed it was hypoxia, which would have triggered far more alarm than our other hypothesis. From this point, the CO hypothesis would have come quickly.

  • As the CO saturation got up over multiple days and was mostly concentrated in our living room, the residents who had been in the house the most had 5x higher CO saturation than those who had been there only to sleep. This made some residents feel very ill while me and others were feeling fine, which made it harder to realize it was because of something happening in the house.

• Eventually, we figured it out because five of us were in our living room and shared that they were feeling off in various ways, and that created common knowledge of the correlated issues. We immediately concluded there must be a common cause, started brainstorming, calling emergency services, and quickly figured out it was carbon monoxide.

Here are the cost we incurred because of this accident:

• One day of lost work and wages for 8 people

• Four weeks of not having hot water, while we waited for an inspection that would allow the gas provider to provider us gas again.

• For me, a ~100€ hospital bill, because I did not have my social security card on me and failed to get through the process of getting reimbursement

So, some cost, but it could have been much worse. If we had not been in the same room this morning, there was some risk we might have taken until the evening to notice, and there was some low risk someone would have fallen unconcious in their room and died in the meantime.

I could not feel safe anymore

The words update was feeling like I was much less safe than before. It was weak, just a bit more of anxiety, of worry, especially when inside our house, but it did decrease my quality of life. I had been suprised by the accident, and higher anxiety was a way to be readier for a world where the rate of surprise encounters with death was higher than I expected before.

The way out was to process the issue, to figure out what I had done wrong, so I could reliably avoid this class of issue in the future. I did an early version of this postmortem, through conversations and notes, until I trusted that my future would not involve more near death encounters than I expected before the accident.

I think my other flatmates also went through this process in their own way. Camille through writing the bulk of Basics of How Not to Die, Elisa through writing her testament.

My updates

Looking back over all the causal chain, here’s the generalized actions I think me and my flatmates could have taken to avoid this outcome.

• Calibrating on which tail risks we were actually exposed to, both through knowing population base rates and specifics of our situation, and taking cheap opportunities to protect against those

• Avoiding living spaces where landlords care more about saving money than the safety of the residents

• Doing our own evaluation of critical systems like heaters. Trust, but verify

• Training in disagreeableness, to feel comfortable pushing back against the landlord when they were prioritizing their profit above our safety

• Learning more about our biology and which symptoms are indicative of which dangerous condition

• Learning more about the other residents, to notice which behavior are within bounds and which are indicative of an issue

• Proactively raising potential issues, be they about the house or about one’s health state, to build common knowledge and notice patterns. Better to have some noise than miss a critical info.

I’m not sure which ones I would have actually taken. All of them come with tradeoffs, costs in time and money that might not be worth the risk reduction.

At least, I’ll keep them on my mind. Maybe they’ll help me notice, next time I’m taking reasonable choices that bring me ever closer to an accident.

In my mind, I saw myself taking BART out of the airport, all the way to Oakland. The dream skipped over a bus ride north, which left me on a sidewalk. I looked across the street, where Lighthaven was... just there, as it’s always been.

My heart swelling, tears getting to my eyes, I thought “I’m home”.

Although it was just a half dream then, I’ve had this feeling, each time I got back there.

Part of my heart is always there, in California.

And yet, I don’t think it’s time to move there. I’ve rehashed reasons again and again.

But my mind keeps returning there.

The issue is the calling. Why? What is calling me? What for?

is there. also. They’re having a great time. They’re living their best life, doing adventures, writing, playing, meeting new people, exploring. I would like to be there and see the wonder and excitement in Celeste’s face when she sees all the amazing places there, and meets all the amazing people!

I wish for adventure, for the thrill of a life that’s going fast, that’s intense emotionally, where nothing is prepared, but at least I know it’s going to be wild.

Here in Paris, I’m missing role models. I don’t have mentors, I don’t have inspirations. I have peers, true, like my flatmates, but no one who’s so good at a thing I care about that I’m magnetized by them and want to be with them all the time and absorb all their power. There are many people that fill this need for me in the Bay, like Oli or Ray at Lightcone, , . At my current work, I feel like the most competent person in the room, at least in the skills I care about, and while it’s nice being congratulated on my skill, I’d much rather prefer being in an organization where I’d have people to look up to.

I wish for role models, for inspiration, for people who embody the ideals and virtues I want for myself, whose sole existence is lights a fire in my soul, which pushes me to become the best version of myself, to reach up to their level.

Lighthaven is opulence. Lighthaven is a shard of utopia. Lighthaven is the embodiment of ambition. It’s a statement to the world that “so much more is possible”. It’s where I’ve spent the best weeks of my life, and met some of my dearest friends. It’s “israel for autistic people” as the meme says.

In the Bay, they can have nice things, because the community of rationalists is large enough and wealthy enough that they can muster a stupendous amount of man power and funding to make places like Lighthaven happen.

And Lighthaven is only the tip of the iceberg, the most salient point in the rationalist infrastructure and social scene in the Bay. There are so many other infrastructure and community projects, be they the network of group houses, the parents caring for each other’s kids, the secular solstice, Mox, Far Labs, Constellation, community dinners, sequence reading groups, ACX meetups. The density of people in rationalist and AI Safety circles is staggering. The social fabric is deep and fractal. Economies of scale and network effects enable shared infrastructure and community oriented ventures. And having so many people with high-tech salaries makes all this scene flush with money, ready to throw it at projects to make life better for people in the community.

And I know it’s exaggerated, but it still feels like much more than I’ll ever be able to build here in Paris.

I’ve built a lot! Two rationalist/AI Safety group houses, shared AI safety offices, regular ACX meetups. They’re my little shards of utopia. I protect and nurture them, and recruit people to help me grow them and make more, but it’s hard, and it takes so long, and nobody has money or time to spare, nor vision of how things could be so much better.

I wish people around me were richer, had more vision of how good life can be, were ready to pour effort and money into making our little local utopia.

Here in Paris, I kinda know everyone in the extended rationalist sphere. The influx of new people is low. It was much higher in the past. I feel like I’m in a closed community, one that closed the door, that pulled the ladder back up, that closed the acculturation pipeline, the halfway homes. It sometimes feels like a huis-clos, a play with an unchanging set of characters, stuck in the same room, who can’t get out.

I yearn for fresh blood! For people who are discovering LessWrong or EA for the first time, and are excited about joining this world! I want bright-eyed idealists, who shake things up, and don’t accept the status quo, and who are not yet set in their path, who try stuff and make thing weird and exciting again.

Lighthaven always feels like that. Whenever I’m there, it’s filled with interesting people, many with bright ideas about the future, who are excited about explaining their newest thing on a whiteboard for five hours, who ask if I want to start a project with them.

It’s also much easier for me to be excited about things in the Bay, because there’s so many people in the rat sphere. I don’t know a tenth of them! Each new connection is a new potential friend, a new growth opportunity, sometimes even a romantic partner.

I wish for a dynamic and exciting social scene, be it from new peeps joining in, or the scene being so deep I’ve yet to meet everyone. I want a scene that’s not yet settled into a groove, a scene where ideas clash and sparks set ablaze new projects and opportunities!

One day, I’ll make a parallel version of this post, of all the things from Paris I’m longing for when I’m in the Bay.

But for now, I’ll keep visiting Lighthaven in my dreams.

Wishing Paris was just a little bit more like it.

I thought that it’d be easy to continue writing after Inkhaven.

I had written a blog post a day for 30 days! And some of them were great!

But now I feel that the well is running dry.

Because I’m lacking purpose.

Because I’m aimless

If I had a specific goal that I dearly wanted, and where writing about it would help, then writing would come easily!

It happened many times before.

Sometimes, I need to process an issue in my life, so write about it to get my thoughts in order and to have a way to transfer context to my friends who could help.

Sometimes I have an idea of something that could exist in the world, and for it to happen, I need to convince specific people that this is an idea worth taking seriously and working on. Then, I write a memo, a persuasive piece, to show them why this makes sense.

Sometimes, I have something to protect, something that I’m afraid will disappear or will never happen, and I know I can’t fix it by myself right now, so I write as a way to scream for help, a way to shake others and tell them to wake up, a way to distill why something is important, so my future self can keep following it even when the heat of the insight is long past.

But right now, I have none of those.

Oh right, I do have some goals that feel worth working on, like getting back to comfortable savings, getting more productive on solo projects, investing in my local relationships and getting allies, getting good at writing and keeping to it to express my ideas and expand my reach. But all of those are instrumental goals to make me better prepared.

Prepared for what?

What am I doing all this work for?

Two years ago the answer would have been easy. It would have been x-risk. The future could be amazing, or we could all die. What else could be more important‽

But now, I feel it much less.

…

Wait… I still feel strongly about not wanting to die! I cried so fucking much after the Solstice because I felt it then, the oncoming doom, the narrow path to a good future, the fact that I was wasting my life away not working on it.

And then, what felt right to do was... To strategize about xrisk until a personal strategy comes out, that feels both ambitious and achievable.

I’m on a waiting loop, improving productivity, getting savings, investing in relations and network. All of those are just generally instrumental, so I’m doing them while waiting for an answer of where to go.

But I’m not working on figuring out where to go! What have I been thinking‽

Finding an ambitious path should be my priority.

That’s it? All I needed was to look at the discomfort around writing for a few minutes, and realize I had the answer since Solstice? I just needed to remember it?

Sometimes, you just need to think by the clock it seems.

If you act like a rich person, you’ll become rich

- paraphrased from

In 2025, I acted like a rich person. The year before had been my most profitable to date. I had not spent a dime of my MATS stipend yet. Opportunities had kept coming my way. It seemed that money was basically a solved issue, I would just keep doing good work in AI safety, and money would find its way to me.

I went on a travel spree, visiting friends in all corners of Europe, signing up for all the expensive camps, all the conferences1. I even went to Inkhaven! I made so many friends and professional connection, and I grew so much! My travel budget tripled.

I moved to a new apartment, right in the center of Paris. It’s much more beautiful, better located, with better flatmates. It made my life so much better! And it’s 40% more expensive than my previous place

I splurged on improving my life: buying cool barefoot leather boots, basically remaking my entire wardrobe to be gothic, getting tons of books and furniture for my new place.

Finally, I did not stress about money. I could just go to the restaurant if I wanted to. I could buy something that struck my fancy in a store, be it an ice cream or a jacket. I developed the TAP of noticing problems that could be fixed by buying something and immediately opening Amazon on my phone to order it.

I knew in the back of my head that this rate was unsustainable given my non-existent income, but I trusted that opportunities to make money would find their way to me in due time. In fact, they did multiple time! But I’d not been tracking my budget, so I did not know where I stood financially. That’s how, in December, I started making spreadsheets.

Modeling

If you can not measure it, you can not improve it.

- Lord Kelvin (apocryphal)

I’m a spreadsheet altruist. Making elaborate models in Google sheets to take decisions is part of my religion. So, in late December, I got to work on figuring out what actually happened to my finances during this crazy year, and what I expected to happen next year.

First, I computed how much of my savings I had used. Easy! Just take the last wealth snapshot of 2024, make a new one for late 2025, compare the two, and... -26k euros.

Shit. Let’s double-check.

facepalm. I had not counted outstanding invoices for 2025, and included the 2024 income tax.

Without those distortions, that’s -10k. Not great, but about what I expected from splurging so much while working less than half the year.

Next was modeling the upcoming year. How much income do I expect to get? How many expenses? Am I on the right track to, you know, have my wealth go up this year instead of down?

I made an expense scenario, one where I basically spend as much as last year, with the fancy new apartment rent, transatlantic flights, camps, restaurants, cryonics, crazy french tax, and... shit my student loan. Good thing I caught this one.

On the other side, expected income. I put my existing jobs, with their hourly rate, hours per week, expected number of weeks worked, conversion rates, self-employment taxes. Each total was converted into a number of months of expenses they would afford me.

And in total, I expected to get enough money to live... 7 months. DAMN.

Let’s back up. This is a wild expense scenario. What about a more reasonable one, with less crazy camps and transatlantic flights? It would be enough to live... 9 months. Double damn.

In my models, I’m expecting to finish the works I’m currently doing by the middle of the year, so I can expect to find about as good opportunities in the second semester, or event better. So, I can tentatively expect to have another job in the back half of the year that bridges the gap.

Although, I kind of hoped that my current jobs would carry me to the end of the year, and that I could travel around in Q3 and work on personal projects in Q4.

Anyway, my models are clear, I’m not on the right path to be rich. I can afford my lifestyle if I stop having time off work, but I won’t get back what I took out of my savings, and I definitely won’t be significantly increasing my wealth. At least, not without some drastic changes.

Mediocrity

I’m underemployed, and I like it. I work 25h per week on stuff that’s quite fun and not too challenging. I have time to invest into my friendships and my social circle, I’m making progress on lots of personal stuff2, and I have ample time to read Worth the Candle. I’m even keeping up with publishing on Substack!

And yet, 25h/week is not enough to become rich. It’s not enough to afford the lifestyle I want.

It’s uncomfortable to stand here, seeing the life I’ve built for myself, and knowing that I’m using a finite amount of savings to keep it afloat. What will I do once I’m out of money? Move out to nowhere? To my parents? Terrifying.

I can’t afford to lose this life. I have something to protect. I need to become stronger richer. Something has to change.

This is what meant when he said acting like a rich person would make me rich. I had to learn how to convert more money into a better life. It took years, and lots of mistakes. Now that I saw all the value I get out of having this much spending power, I know how much of a hit to my quality of life it would be to be poorer.

No need to hit myself with an ought here. I feel it in my bone, the desire to be richer. It’s ever present. It’s keeping me up at night. It makes me think and think and ruminate and plan and talk and ruminate and try stuff out until something sticks and a path towards greater income shows itself.

Part fear of losing what I built. Part hope to continue on my trajectory.

I look at my current plan, and I think this is not enough. This job is mediocre. I’m worth more than that. I can have much more impact, and be much better paid elsewhere. I know my services are valuable, be they event planning, programming, or strategic advice. I could find much better opportunities than what I have now.

And yes, this would require doing more work, maybe even working 40h/week, or even more. I know I can do it. I’ve done 70h/week before, and enjoyed every minute of it. I don’t need to settle for a mere 20h/week main job, nor for half of the max rate I’ve ever been paid. Combining a base job with extra hustle with better pay and more hours was part of this, but I need to find something that scales to more of my time and pays at least 30% more.

And yes, this would require going on a search for better jobs, filling out applications, training by myself, networking. I’ve done it before, and that’s how I keep receiving opportunities even though I did none of the search work. As per the Gale-Shapley algorithm, if I wait for employers to make me offers, I’ll get the worst possible offer, whereas if I go seek out the best job I can, I’ll get the best possible offer (or something like that). Applying to Lightcone was part of that. Now, time to do more.

And yes, this would require doing challenging work, where I’m not sure if I’ll even succeed. Settling into mediocrity is great for being relaxed, but a terrible way to increase my skills and agency. I want a job that keeps me on my toes, that pushes me to become a better version of myself! Going to Inkhaven was part of this, but it’s not enough. I want a job that’s so important and so far outside my skill range that I have to get my shit together and learn as fast as I can to have a chance. That’s what I did in 2024, and I became fucking great. The only way to get there is to apply to jobs I’m woefully underqualified for, get accepted, and scramble to succeed at it anyway.

I want to say that I won’t settle for mediocrity, that I will become stronger, for impact, for money, and for the love of the game. The issue is, I already thought that in 2025, and it did not really lead there. 2026 will have to be different, because I’m on a limited runway.

Will I scale down my life and settle for a chill life, or rekindle my ambition?

So little time.

An incessant stream of tasks, of messages, of projects, of meetings.

I knew what I was getting into, but I’m still surprised at the strength of the current.

I am still swimming.

My schedule is tightening. My agenda is full. I tell my friends that we can meet some time next week. Next is already filling up alarmingly fast.

I am still swimming

Goals have been tuned, clarified, sharpened. I know where I’m going. It’s just a matter of execution.

I am still swimming.

I prepared for a whole month for this.

I am still swimming.

I know that what will keep me afloat is my systems, having good sleep, going to the office every day, not letting inboxes fill up, and saying no to keep at 80% capacity.

I am still swimming.

I did not write for five days. First a rationality retreat, where I used most of my Beeminder buffer. Then two days of working all day, with socials in the evening. I did not find the time.

I am still swimming.

There is now work, and work can fill up any amount of time if I let it. There is always more to do. And when this is the case, it’s hard to focus on personal stuff, like writing, going to the doctor, fixing issues at the group house, doing any of the myriad of personal projects that I’ve been dreaming of doing.

I am still swimming.

I’m already seeing my Toggl fill up with the hours I’m working. I’m already imagining the first invoices I’ll send, at the end of the month. My wealth has been draining away for too long. Now, it’s time to save again. This was the right choice, but man does it hurt to close off the other paths, the ones where I work on what I want, the ones where I focus on writing, on making friends, on improving my life.

I am still swimming.

But for how long? The walls feel so close. 80% capacity I said, that’s the maximum. I negotiated to not work on Friday, that will help. 4 days a week is already 80%! I got a full day for all the other work. I just need to protect this time, and not burn myself so much on the rest of the days that all I can do on Friday is recover.

I am still swimming.

But will I swim still? The jobs feel like they’re pushing me, but I’m not sure if they’re pushing me in the right direction. I am continuously improving at focus, efficiency, and the skills of being an executive and thinking assistant. But is it pushing me forward enough? I feel like I’m honing my skills, but not learning new ones.

I am still swimming.

The current is strong. Last time I was there, the current started pushing me backward, towards the precipice. The panic left me scrambling, desperation ruining my form, leading me to move backward even faster. This time I’m more prepared. I see the treasure at the end, just need to swim further. Focus on the movements. Arms, legs, breath. I have the energy, I have the technique. I just need to execute.

I will keep swimming.