I am taking stock of my first year on Substack, and my first year with the resources necessary to focus on attempting to be a public intellectual of sorts.

The results are a mixed bag. Things are progressing, but slowly. Everything takes longer than one thinks or expects. Finding useful help has proven difficult, although I think I have found the right person and she should be able to start soon. Growth in reach has been similarly slow.

My biggest disappointment is that I have not done as much long term or evergreen work as I would have liked. I haven’t laid down enough building blocks, progressed the codifying of my intellectual models, the way I need to if I want to meet my long term goals.

This became more clear as I looked back. I wrote quite a lot in 2022, and am happy with the quality. Yet so much of it is clearly ephemeral.

A lot of that, looking back, is length. I wrote long posts and did not take the time to write shorter ones. If I want to build something over time, I need to take the time, at some point, to write shorter ones. Perhaps they are summaries and glossaries. A Wiki or Roam of sorts might be involved. It needs to be something.

What will I be able to best carry forward into 2023? Which posts accomplished something in the longer term? What do I make of them now?

To apply the message ‘summary needs to exist’ if I was looking back and telling people what they should read and remember, I think these are the top eight most important going forward, in order (again, excluding Balsa), only 3 of which were in the top 10 posts by hits:

  1. On Car Seats as Contraception
  2. On Bounded Distrust
  3. The Long Long Covid Post
  4. Repeal the Foreign Dredge Act of 1906
  5. Key Mostly Outward-Facing Facts From the Story of VaccinateCA
  6. How to Best Use Twitter
  7. Formula for a Shortage
  8. Grand Theft Education

Here are core lessons I’m taking away from the experience, including some things that don’t come up in the discussions below.

  1. Focused posts have more upside.
  2. Draw a sharp contrast between speed premium posts and non-SP posts.
  3. Look for posts that should exist, where no one else would write them.
  4. Gaming content hurts the blog, needs to be moved elsewhere.
  5. Subscription revenue will never be my core business model.
  6. Metrics on individual posts are probably better than nothing, but it’s unclear.
  7. Everything takes longer than you think.
  8. Better hardware is a big win. Work with multiple huge monitors on new desktop.
  9. No one clicks on links, maybe ~25% of users click even one in a giant post.
  10. Only path forward is to do what seems right to do, see where it leads.

The rest of this post looks back at the top 10 posts by traffic, plus any others that seem important in the long run, excluding the Balsa posts, then talks a bit about what didn’t get written and didn’t work.

A central theme of many of these posts is: I wish I had structured them around and talked more about things that would endure, and less about things of the moment.

Yet that is hindsight talking. Such posts would likely have been seen by less people, and been less good at the time. How many people actually go back and read things months or years later? Based on the numbers I have, not many.

Similarly, we say ‘no one on their deathbed ever wished they spent more time at the office,’ except that (1) this is totally not true and plenty of people think this, and more than that (2) it is the person looking back that has the greater distortions in their thinking, and also has been severely de-risked.

Top Ten By Hits

#1: On Car Seats as Contraception

This post seemed to hit quite a nerve.

I’ve been strongly pro-natalist for a while. People are good, and conditions are better when the population is growing and worse when it is aging and shrinking. The world would benefit from having more people.

While there is the potential for AGI or other technological changes to alter our course or save us from the trend of falling fertility, it would be a poor strategy indeed to rely upon such a rescue.

I had been mulling over the original paper on car seats for a while. Car seats are an example of a class of requirements that together make the effective cost of raising children much higher than it needs to be, while offering little if any benefit in return. If it was possible to have a sane approach to child supervision and education, and to not obsess over every little thing and have to face endless costs while having no free time or flexibility, people (including my own family) would be willing and eager to have more kids.

The direct motivation to write the post came when I realized that I could use the data to back out an approximation of how much it would cost to induce additional births. That number turned out to be ~$270k, approximately the cost to raise a child in America.

The article, in turn, was one of the bigger motivating factors in moving forward with Balsa Research, likely second only to Repeal the Foreign Dredge Act of 1906.

There were four essential classes of important pushback.

The first class of pushback was arguing that the seats were indeed an Important Safety Feature. My answer to this is: No Just No. One can quibble with the exact numbers. It’s never going to be the right order of magnitude to justify the costs and negative effects.

It also occurs to me that any argument for mandatory car seats is also a much stronger argument for outright banning cars, which have always been an unprincipled exception to our rules about safety. Our continued existence depends on our ability to support such unprincipled exceptions, and I worry we are losing such ability over time.

The second class of pushback was a physical world argument that one could, actually, fit three car seats into a normal car, if you put your mind to it. This would still be super annoying even after implementation. You’d effectively only be able to travel in your own vehicle, which would not be good for many other uses. Getting the kids in and out would be a substantial pain. It would still seem not to be the same order of magnitude of bad, if it worked, compared to having to buy a new car.

My main response to this is that trying to do this is unusual and weird, it definitely has its own problems, and in practice the vast majority of people never consider the possibility or attempt to make it work. In practice, that is all it takes for the tax to be real and births to be discouraged. It would still be valuable to spread the word more that this is an option.

The third class of pushback was to call the whole thing ridiculous. There was no way that someone would not have a child due to such a silly little rule, people don’t work that way, come on. That’s a translation into my language, but disdain very much in original.

My response is that no, people really are like this. They make major life decisions based on trivial inconveniences all the time. People would totally not have a child because they don’t want to give up their car, or more so because they can’t afford to replace it. Such constraints often bind, either emotionally, financially or both, and one must always think on the margin. People also read such rules as signals about the way we approach raising children, and correctly intuit that their life experiences will be increasingly sacrificed in the name of the appearance of safety out of fear of social or government retaliation.

The fourth class of pushback was that there was no way something like this could be changed. Once a safety rule is instituted, that rule is forever, it is nearly impossible to take back. No one is going to fight for getting rid of seats meant to protect children. If they did, then the moment there was a fatal accident without a seat, it would be blamed on the repeal, even though it would (1) probably have still been fatal with the car seat, the statistical difference is very small, (2) we won’t know if the mandate would have been adhered to, (3) such fatal accidents are in any case rare and no one involved is doing any of the math. Doesn’t matter, goes this argument. People would get blamed, so you’re screwed, and there’s nothing to be done.

I take this objection seriously. Doing anything about car seat requirements, even though they are useless expensive obvious nonsense, is going to be far more difficult than it would be if it wasn’t a safety issue. It’s not near the top of my list of Balsa’s top priorities to explore. I still think it is worth trying and exploring, in part because most parents I talk to about this have a passionate and violent agreement that things have gone far out of control on such fronts.

One of the original study’s authors responded to the post, excited to explore some issues further. I still have to follow up on that to make sure it happens.

#2: Jailbreaking ChatGPT on Release Day

This was a simple case of ‘this seems like an interesting and important phenomenon that is happening, yet it is all ephemeral and difficult to track, someone should write it down in one place, and I bet no one else will.’

I notice this is a common failure mode. Essentially no one is making such complications. Going forward, I am going to strive to do more posts like this – they are relatively low effort and low time commitment, people seem to get a lot of value short term and they create a permanent reference point.

In addition to being fun and being a lesson in how these particular methods are most easily subverted, the incident is important going forward, and important to document for after these particular exploits get patched, to make clear some combination of:

  1. What currently passes for ‘safety’ is only even trying to be the superficial appearance of an attempt to prevent bad things that stops people via trivial inconvenience, but that will not stop bad actors who care and never will, and would never protect us in any way from an actual AGI.
  2. The work of those who think they have made an AI ‘safe’ or mostly safe usually breaks down within an hour or so of exposure to people who want to defeat those precautions.
  3. The particular technique used here, RLHF, won’t work when it matters. Ever.

#3: Sadly, FTX

Mostly this was me taking my usual default strategy for covering events and applying it to the whole FTX situation. All the news was broken on Twitter while traditional media sources were useless, and I have overlap with the story’s core elements in many different ways, so it was an excellent match.

Effectively, there was a timeline of ‘here are the things I am confident happened, in logical order’ with ‘here are all the things people have said that seem worth knowing about’ and then giving my own takes on it all, so at least three posts in one and plausibly more, it was very long. As with many other things, I don’t see any other one stop shop that attempts to provide the information one actually needs to know about what the hell happened, even without the takes section. This seems like something we are bad at providing.

#4: How to Best Use Twitter

I’d been meaning to write this for a while. A lot of people feel lost on how to approach Twitter, or far worse are doing it wrong. To this day I keep telling people they need to use Tweetdeck, including many that really should already know better.

On reflection it is unfortunate that this post got tied so much into the war in Ukraine. It might be worth an updated version that was more timeless by avoiding that.

What hasn’t changed things much is Musk buying Twitter. We have neither seen our hopes nor our fears realized. I didn’t fear censorship or suspension before and I don’t now. For all the talk of how Twitter has changed, my take is that Twitter has changed very little. The only substantial change is that some accounts have left, which shouldn’t change your approach, and I haven’t lost any that felt important to me.

#5: Book Review: Talent

I don’t read enough books, and I don’t write enough book reviews – I can’t remember ever writing one and being unhappy that I wrote it. Talent was a book full of ideas worth thinking about and discussing, so it was a good match. I found it useful to write the review. It is harder to tell whether it was useful to others.

I had tons more notes, set myself a deadline and stopped when I reached it. There was a second half of this that could have been written, which I did not get around to writing while the topic was fresh, and it’s too late now.

#6: The Long Long Covid Post

This was a long time in the writing, the cumulation of over a year of endless tsouris and arguments. It was also the highest effort post of the year, and one I owed for some time before the debt was finally paid.

I continue to stand by the conclusions and in particular the basic structure and core thinking. If anything, I have updated modestly in favor of Long Covid being less scary. Partly this is based on several recent studies with better controls, making me think I was underestimating the extent to which conditions were pre-existing or coincidental. Partly it is based on continuing not to find anything too scary, a reduction in uncertainty. When you’re dealing with a potential big danger, a lot of the danger is in scenarios that can’t be ruled out but are unlikely, and now seem a lot less likely.

What I regret not emphasizing more is Long Other Infections, and the general tendency of humans to have things mess them up in chronic fashion. We get the impression that the default human state is essentially perfect health and that if you have things wrong with you long term it’s pretty weird. That simply isn’t so. It definitely isn’t true at my age let alone older.

I don’t think the right move is Chronic Condition Acceptance, or pride, or anything. These conditions collectively are quite bad, making lives much worse. We should be working to make people better. We should also of course be working to make people better even when their current state is ‘normal.’ Better is better.

#7: Mandatory Post About Monkeypox

I am happy that I never got all that worried about Monkeypox, and that this was in hindsight mostly for the right reasons. Things were superficially scary for a while. I did hedge my bets a bit, holding out a single-digit probability (at least at first) that things could get more serious. Looking back, that seems reasonable to me.

#8: Convoy Crackdown

Most of a year has gone by. Occasionally someone goes ‘remember that time the Canadian government froze the bank accounts of the family members of people who contributed money to a protest against Covid restrictions?’ and the answer will be no, we do not remember that, nothing important happened today. Everyone went back to their business as usual.

Does that mean that the concerns were overblown? Or does it mean the concerns were fully realized?

Do you feelfree to transact?

#9: Paper is True

The fundamental question is, if those on one side lack analytical abilities or are more generally outgroup members who hold Wrong Views that conflict with Narrative, to what extent should we run the other way? Should we treat this as evidence that such people are wrong, or that we should avoid association with them even if they are right?

This problem gets worse when we consider the increasing alignment into factions, one of which reinforces Narrative and supports The Current Thing, one of which opposes them, and of course neither of which has much in the way of strong analytical abilities.

One could also go to the second level. If those who have strong analytical abilities have concluded that they should not openly endorse something, perhaps you should trust that they have concluded this for a good reason, even if the thing is true or desirable. If all the smartest people are clicking the ‘collaborate’ button, shouldn’t you?

I hope you already know my answers.

#10: Covid 6/30/22: Vaccine Update Update

This was the only Covid post in the top 20, and the only one to get substantially more than the current median number of views. This was, as I suspected, because of a link. Reading it over, it has some good information and analysis in it, but nothing out of the ordinary.

Repeal the Foreign Dredge Act of 1906

The Jones Act is the ultimate symbol of harmful protectionist legislation. The Foreign Dredge Act is still remarkably impactful, possibly more impactful, and the case is that much cleaner and more obvious while the private interests that benefit are much smaller. So I started there.

Since then, the focus of those who focus on such things has been squarely on the original Jones Act. With the existence proof of being able to make such a case in hand, I was led to create Balsa, and there too the pull is towards a Jones Act focus rather than a Dredge Act focus.

I do wonder if that would be a mistake, though. There really is a much easier deal to be made on the Dredge Act, where only the direct corporate beneficiaries – the shipyards and the owners of the existing ships – benefit at all, with the magnitudes small enough that cutting them checks is practical. Every single employee on those ships would still be employed, if anything more often on better terms.

Buying the losers off might look bad, but look at the omnibus bill currently going through Congress and ask how high up the looks bad chain it would get if it had been stuck into there.

I continue to think this is a sleeper issue of far more importance than people realize.

The issue is more that any given person advocating for this has to either advocate only for this repeal and not the Jones Act repeal, and be obviously doing that because of how it looks, or else advocate for both and be seen as targeting the Dredge Act as a mere stepping stone. Not sure what to do about that.

On Bounded Distrust

Rereading made it very clear: This was too long.

That doesn’t mean there were clear wasted sections. One still has to kill darlings and find a way to say the important parts of this faster. Or at least write a summary. Asking someone to Read the Whole Thing is not reasonable.

That is a shame. Getting a good sense of how Bounded Distrust works is a necessary requirement to navigating the modern informational environment and internet. That no simple explanation and guide yet exist is a sign of how unreasonable this is as something to expect of everyone in a society.

I am declaring that I owe a shorter post that gives practical advice on how to apply bounded distrust. The Urgency of Normal does not work here, we don’t have time to work through an example, this needs to be an explicit ‘this is how this works and here’s what to do with that information,’ ideally in 2000 words or less.

On the Diplomacy AI

This got the man who built the AI to show up in the comments to challenge my statement that I disliked the AI’s tactical play, saying it was was world class at Diplomacy tactics and gunboat diplomacy, and that I was wrong in particular about the example I pointed to, leading to a detailed discussion, an appeal to expert opinion, and me playing a gunboat game against some bots with exactly the amount of lag between turns that drives one crazy and also the experience of playing Germany and having Italy keep trying to swipe your centers, to remind me how infuriating and stressful the game can be.

While I didn’t retract as such, I did modify the post to include the other side of the story. On reflection, I think that I was holding the AI to too high a standard, more than anything else. I was miles better at tactics than the people I ever played with. It wouldn’t be that weird if I was one of the best players in the world at tactical Diplomacy back when I was getting the practice, so the AI could still be getting large net benefits on that front while not impressing me all that much.

Key Mostly Outward-Facing Facts From the Story of VaccinateCA

The story of VaccinateCA is the story of a group that came together to create a website and data source. It is also the story of our vaccination efforts, in all its insanity, incompetence, perversity and failure, all of which is continuously being pushed towards a memory hole. Both halves are a hell of a thing.

I wrote the summary because Patrick’s long post is long, and it is important to have a place to point people and to reference that is shorter and focuses on the things it is most important to remember did indeed happen. I also wrote it to help ensure more people read the original, and to highlight places in which Patrick was, shall we say, burying the lede or being polite, and the important points need to be made text.

The key difficulty with response/summary/reaction posts like this is to ensure that they are written such that they will endure over time rather than ending up ephemeral. I think I did that here. Hard to tell.

Formula for a Shortage

The hope is that going into detail about such things will provide examples that help us understand such dynamics for next time. Or at least that one now Has The Receipts for the incident, which can be part of your Having The Receipts in general when the problems inevitably surface somewhere else. Will it matter? I would hope so. At least a little.

Grand Theft Education

The Biden Administration’s student loan forgiveness plan was truly terrible. We rewarded those who took on debt and didn’t pay it back, while mostly encouraging the predatory system that forced those loans onto students in the first place. There were places judged so terrible all outstanding loans involving them were forgiven, and we didn’t even ensure they won’t get any similar future loans.

Then there’s the income dependent repayment plans, which are essentially an invitation for students to take on truly massive amounts of debt, then not earn much money, and then never pay the loans back. And for graduate schools in particular to massively hike tuition in order to then borrow more government money and then not pay it back. Or they could simply take out massive loans regardless, the money does not need to be spent on tuition. It can be put into a savings account, go ahead, it’s fine.

The whole thing is of questionable legality, which might save us the over a trillion dollars this would cost, stolen from public funds to give to one’s political allies.

When I think about it, this all still enrages me. People should remember. People, who are not getting free money from the rest of us, should care about this.

Those people, alas, do not seem to care about this. It does not land, or endure in memory. I need to better understand why this is so, and what might be done about it.

Against Active Shooter Drills

My younger son’s school does these active shooter drills. Periodically they take time out of his day to traumatize him. It is not good. When I told them I did not want this, they replied that it was mandated by state law. When I asked to be warned about future, I was told that state law would not allow that either. It is mandatory that my child be placed in school, where they will intentionally traumatize him. It is further mandatory that I not be warned about their intent to traumatize my child, in case I would try to prevent that from happening.

So am I following my own advice on this? I did do the easy steps of trying to prevent the problem. No dice. That only leaves considering other educational options. Which, yes. We are considering, but there are many other large considerations at play and none of those options are good. As terrible as the situation with drills might be, there are so many other terrible problems that are bigger.

Le sigh.

On AGI Ruin: A List of Lethalities

This was my first and so far only commissioned post. I was doubtful my thoughts were worthwhile, multiple people disagreed enough to pony up the cash. I hope that they were right, and I think they were right that this was a good use of time. Certainly writing it helped clarify my own thinking, and that of a few people I know who I discussed it with.

As with Bounded Distrust, I notice that for this to be more long-term useful it would need (or at least want) to have an easier way in, and allow people to get its takeaways faster. I continue to be unsatisfied with the availability of a simple, quick explanation for why AGI likely destroys all value in the universe and AI Dontkilleveryoneism is an important cause.

In an ideal world, the author of that post is most definitely absolutely not me. That doesn’t rule out that I should take a crack at writing it anyway at some point.

What is Conspicuously Missing?

What either conspicuously didn’t get written, or is clearly not working?

I made a decision not to fully cover a number of events. Most of them don’t stand out.

One that does is only doing one post on Elon Musk and Twitter, then bowing out. I have a bunch of draft material on this, I could have kept going, and I chose not to, despite there being some important stuff here. On reflection, I regret not pushing forward at the time. I note I should double back some time soon. I’m not quite going to say I owe a Musk post, but I will say I need to strongly consider doing it.

I am happy with my decision to do early Ukraine war coverage, and with my decision to stop covering it once the situation became slower moving and more clear. If there were ten of me it would make sense to have someone on that beat.

The biggest thing missing was, as I mentioned at the top, more shorter or longer term posts designed to build up a structure over time. I need to fix this in the new year.

What Failed?

I tried launching a discord server. It turned out not to be worth the investment necessary to generate momentum. I’m happy to allow it to exist, and I think what conversations do happen there are good things. The problem is that one only has so much budget for such places and actions, and the bar is high. For now, no good.

There was one post, on an RCT of colonoscopies, that I quickly retracted and unpublished. There were a bunch of distinct things that went wrong here. I didn’t sufficiently strongly seek out enough feedback from the right people given the weight of the issue, the implications of the conclusions that seemed right, and my lack of detailed knowledge. I interpreted a lack of anyone pointing out things as being too strong evidence for those things not being there to point out, and didn’t sufficiently check. Among other things. There’s no way to salvage the analysis other than ‘do a completely new analysis’ and the default is that I don’t end up doing it given potential other topics.

If anything, I am suspicious of the extent to which I continue to endorse my previous thoughts and opinions on reflection as I reread the year’s work. I worry I am not often enough changing my mind, on matters both great and small.

I continue to play games and I continue to write reviews of those games and other gaming thoughts on occasion. Although gamers often express appreciation and find my thoughts interesting or useful, these posts reliably cost subscribers and generate relatively little interest.

Thus, I’ve concluded that such posts no longer belong directly in my Substack. The choices are to create an additional Substack, or to simply leave them on WordPress, and offer links in my roundups. An advantage of a new substack is that it could be a free place to offer various gaming thoughts with very low activation threshold, since it wouldn’t be in the face of anyone who wasn’t interested. So that’s the way I’m leaning.

This isn’t the topic of this post so I won’t go into detail, but something else that didn’t work was hoping to pay the bills via subscription revenue. While there is gradual steady growth in free subscriptions and readers, subscription revenue is flat at a low level – the paid subscriptions I do get are helpful and appreciated, thank you to everyone, yet that is not going to be the primary way my writing and this blog are funded any time soon.

I also don’t discuss Balsa in detail here. I’m waiting for certain technical ducks to be in order, in particular finishing setting up my pass-through for donations – we’re happy to accept donations and you can contact me for bank account and other payment information, but what we can’t do is assume you that it will be tax deductible, which many people need. I figure once that’s done I will do a ‘here is where we are and how you should think about whether to donate’ post, and go from there.

New Comment
8 comments, sorted by Click to highlight new comments since:

Yet so much of it is clearly ephemeral.

This is also my main criticism of your posts. As of late I mostly try to avoid them for the same reasons I avoid the news. In some ways they are actually worse for me than the news because I agree so hard with them and they trigger me with "look how irrational people are" bait.

No one clicks on links, maybe ~25% of users click even one in a giant post.

Two comments, with detail below: (1) make sure you have the relevant denominator and (2) be careful about taking action based on this information.

(1) What counts as a user in this context? Someone who comes to the page, reads a sentence, and then closes the page wouldn't even have time to click a link, for example, but they don't represent who your readership actually is. Similarly, users can end up double counted where, for example, they read through the post on their phone, and then come back on their computer to copy a quote or comment. I expect the relative numbers of link-clicking to be a useful comparator between blogs, but I'm not sure how to make sense of that number in a vacuum.

(2) Supposing this is basically true, does it change how you want to write? I expect it depends on who you are writing for, but I predict that the quality of your readership would go down if you didn't link to your sources.

I continue to be unsatisfied with the availability of a simple, quick explanation for why AGI likely destroys all value in the universe and AI Dontkilleveryoneism is an important cause.

In an ideal world, the author of that post is most definitely absolutely not me. That doesn’t rule out that I should take a crack at writing it anyway at some point.

I’m working on this atm! I also find it surprising that it doesn’t seem to exist yet.

#8: Convoy Crackdown

Does anyone know how long the bank accounts were kept frozen? Are they still frozen? Did it only last a week?

Update: According to wikipedia, the use of the 'emergency act' to freeze people's accounts was revoked 9 days after it was initiated, and 2 days after Zvi's post. I don't know with confidence that the people's accounts ceased to be frozen after that point, and I'm sure it will continue to be on their long-term financial records, but hopefully it means the freeze was lifted soon after and not left indefinitely.

The link doesn't go to Wikipedia.

Thx, fixed.

I’m curious what fraction of high-income people have kids with student loan debt that’s been forgiven. It may be that the people who’ve lost the most money on net to forgiving other people’s student debt are simply a tiny fraction of extremely wealthy people and people with no close relationships with student debt holders. If so, it’s no surprise if most people are fine with the program. This is all speculation, and if anyone has data on this I’d be interested to know the answer.

TYPOs

no one is making such complications

Compilations?

Do you feelfree to

missing space