rachofsunshine

To where are you local, what are your desiderata, and what does your resume look like?

I can take a look privately if you'd like, or publicly here if you want broader opinions / to serve as a data point for others.

Rationality on an individual level is not the same thing as what produces the best long-term outcomes for both parties. On an individual level, bringing up comp immediately significantly reduces your chances of being moved forward. It shouldn't, but it does.

See this other post from us: https://www.otherbranch.com/shared/blog/would-you-still-hire...

This is a specific application of a good general principle. Big companies need to watch for failure modes. Small ones need to watch for success modes, because the default is always failing.

We can. That's kind of the entire point of our business. Check back in a week or two - we've got another blog post on some of our interviewing data in the pipeline.

Yeah, so did I. Being both a ride-or-die leftist and the owner of a company is a weird place to be sometimes, and it's basically the way I figured I could best implement the world I want to see inside the world we have.

It isn't, but neither is the original post! It's an important addition.

Those things are fakeable, but there are plenty of people who will aggressively signal a LACK of hunger. It's more of a negative predictor than a positive one.

Yeah, to some extent you have to be willing to deal with this stuff in recruiting, which is why I've taken on the clients under discussion at all.

In my Triplebyte postmortem (also on the blog), one of the mistakes I talked about was that Triplebyte was aggressive about trying to dictate terms. We told people how they had to hire.

Otherbranch takes a softer approach: if you ask for my opinion, I'll tell you what I think. Otherwise, I'll do my best to find you what you asked us for, with the understanding that some sets of constraints reduce the probability of success to ~zero.

That goes on the candidate side, too. I get a fair number of people who will come in and tell me "I only want a remote job where I can take a day off whenever I want and only want to work on a super clean codebase and also get paid 250k a year" - and those people are almost never going to end up with jobs. But the tradeoffs they want to make are their business, not mine, until they ask me to do otherwise.

It doesn't even need to be "good enough". People SHOULD be picky about founding engineers. But they should be picky about HIRES, not about top-of-funnel proxies for skill.

This is a good addendum. Do you mind if I add it to the post (credited, of course)?

Depends on the business.

Some startups (like mine) are delivering a service, and the technology used to deliver that service is instrumental. Our back-end is an Airtable I configured myself, and it's been sufficient so far; better tech is not make or break for what we do. Other startups, like Flexport some years ago, fundamentally depend on technical function because that's the core of what they do.

One of the common mistakes founders make, in my expetience, is not asking which camp they're in. It's not a hard question to answer (usually), but it's an easy one not to ask.

There is some truth to this, but I would argue (with a considerable amount of data on both assessments and hiring behaviors) that it is less true than people might like to hope it is.

I very intentionally did not write anything about finding engineers who are just good at the things you care about and not at other stuff, because every bit of data I have says there is a considerable component of general engineering skill underlying most eng roles. No, it isn't totally one dimensional, but (in a principal-component-analysis sense) it is fairly low-dimensional.

There really are just better and worse engineers in the sense that eng A is better than eng B for virtually every job. But that's precisely why recognizing the competitiveness of hiring is important - the more you insist on narrowing your pool, especially in ways others also narrow theirs, the less likely you are to find the rare unknown great engineer.

There's a bit of doublethink involved.

On the one hand (and as I mentioned in the post), yes, most employers are not as dumb as I'm making them sound. In principle they know they need to comprpmise - but in practice, they often balk at doing so because they haven't clearly articulated what they will compromise on.

And, more to the point, that they'd hire a _better_ imperfect candidate by taking those four months doing tough interviews with lots of imperfect candidates (rather than hiring one in desperation later).

The market is definitely not dead. It started warming up last summer and has continued to do so throughout 2025.

But the market is two-tiered in a way it hasn't been before, particularly w.r.t. remote hiring. Almost all engineers want remote jobs and a small number of employers offer them, so the remote job hunt still puts employers in the driver's seat. But (good, senior) engineers hold the cards right now for in-office roles.

I've been in recruiting for seven years! If I weren't frustrated with clients sometimes, there'd be something deeply wrong with me :)

Author of the OP here - to put some more empirical backing to this, virtually every single engineer in our candidate pool values illiquid equity at 20% or less of face value, and about one in three give it no weight at all.

Totally off the topic of the thread, but it's why I do things differently with the people who work for me. I'm the sole owner of Otherbranch, but I pay out a percentage of profits over certain thresholds (between 25 and 75%, rising at higher levels of profit) to the team. Keeps things concrete and aligns incentives with building something that works today rather than obsessing over a hypothetical exit.

Sometimes simple points are important, especially when those simple points are not generally agreed-upon. A proposed law declaring pi to be 3.00 would indeed be bad, and the simplicity of its badness would not make it any less bad.

Zero-knowledge proofs allow unique consumable tokens that don't reveal the individual who holds them. I believe Ecosia already uses this approach (though I can't speak to its cryptographic security).

That, to me, seems like it could be the foundation of a new web. Something like:

* User-agent sends request for such-and-such a URL.

* Server says "okay, that'll be 5 tokens for our computational resources please".

* User decides, either automatically or not, whether to pay the 5 tokens. If they do, they submit a request with the tokens attached.

* Server responds.

People have been trying to get this sort of thing to work for years, but there's never been an incentive to make such a fundamental change to the way the internet operates. Maybe we're approaching the point where there is one.

Everyone's just playing a bunch of investor signaling games, because founders/VCs by nature are living in the world of finance, not the world of decision-making (at least in their public personas). It's part of why I opted out of that whole world when founding my own company. Not literally so that I can listen to my angsty 13-year-old Christian rock without answering to anyone (although I certainly do do that), but because I feel like that fear colors a whole lot of what gets done in tech these days.

Engineers very often tell me something like "well I have this idea but I don't think anyone will fund it" and - well, just build it, man! Your idea takes like two grand of startup capital, and I know for a fact you made 240k last year. There's this whole mythologized idea of founders as a separate breed, encouraged in no small amount by founders themselves, but...founding a company is literally just building a thing people want and selling it to them. You can wear clown shoes and do that.

Should we trust the information at face value without verifying from other sources? Of course not, that's part of the learning process.

People who are learning a new topic are precisely the people least able to do this.

A friend of mine used chatgpt to try to learn calculus. It gave her an example...with constants changed in such a way that the problem was completely different (in the way that 1/x^2 is a totally different integration problem than 1/(x^2 + 1)). It then proceeded to work the problem incorrectly (ironically enough, in exactly the way that I'd expect a calculus student who doesn't really understand algebra to do it incorrectly), produced a wrong answer, and merrily went on to explain to her how to arrive at that wrong answer.

The last time I tried to use an LLM to analyze a question I didn't know the answer to (analyze a list of states to which I couldn't detect an obvious pattern), it gave me an incorrect answer that (a) did not apply to six of the listed states, (b) DID apply to six states that were NOT listed, even though I asked it for an exclusive property, (c) miscounted the elements of the list, and (d) provided no less than eight consecutive completely-false explanations on followup, only four of which it caught itself, before finally giving up.

I'm all for expanding your horizons and having new interfaces to information, but reliability is especially important when you're learning (because otherwise you build on broken foundations). If it fails at problems this simple, I certainly don't trust it to teach me anything in fields where I can't easily dissect bullshit. In principle, I don't think it's impossible for AI to get there; in practice, it doesn't seem to be.

New networks always start on top of old ones - presumably someone will, at some point, standardize an agent interface for these sorts of systems.

Funnily enough, we already have a precedent for computers communicating by phone: the modem! The more things change...

What makes Goodhart's Law so interesting is that you transition smoothly between two entirely-different problems the more strongly people want to optimize for your metric.

One is a measurement problem, a statement about the world as it is: an engineer who can finish such-and-such many steps of this coding task in such-and-such time has such-and-such chance of getting hired. The thing you're measuring isn't running away from you or trying to hide itself, because facts aren't conscious agents with the goal of misleading you. Measurement problems are problems of statistics and optimization, and their goal is a function f: states -> predictions. Your problems are usually problems of inputs, not problems of mathematics.

But the larger you get, and the more valuable gaming your test is, the more you leave that measurement problem and find an adversarial problem. Adversarial problems are at least as difficult as your adversary is intelligent, and they can sometimes be even worse by making your adversary the invisible hand of the market. You don't live in the world of gradient descent anymore, because the landscape is no longer fixed. You now live in the world of game theory, and your goal is a function f: (state) x (time) x (adversarial capability) x (history of your function f) -> predictions.

It's that last, recursive bit that really makes adversarial problems brutal. Very simple functions can rapidly result in extremely deep chaotic dynamics once you allow even the slightest bit of recursion - even very nice functions like f(x) = 3.5x(1-x) become writhing ergodic masses of confusion.

The private citizen has the option to pay cash, bitcoin, trade goods or services, etc...everyone is[n't] entitled to convenience in every situation.

These options aren't small impositions, they're sufficient added overhead that they dwarf the value of the transaction itself. Bitcoin is the only one of them that seems vaguely realistic to me, but most people don't (and shouldn't) keep their own crypto wallets and don't (and can't) get paid in crypto, so that still requires interaction with third-party processors on two levels. It needs one level to convert fiat to crypto and vice-versa and another to conduct the crypto transaction.

Put another way, the sites that are shutting down this content clearly have substantial financial incentive not to do so. If they thought they had a reasonable alternative, don't you think they'd be using it? And if decent-sized companies with financial incentives cannot find an alternative that seems practicable, what makes you think private individuals are reasonably able to do so?

The broader issue here is one of monopoly, and I guess it might be helpful to zoom out here a bit. Do you think a company with market dominance should be able to engage in (otherwise legal) anticompetitive practices to suppress new companies in their domain? If it were up to you whether to have anti-trust law, would you have it?

If yes: isn't this essentially the same problem? These payment processors have a duopoly and are suppressing alternatives who would take these payments (and might outcompete them in the market on that basis).

If not: are you not concerned about a failure-state where monopolies (a) control critical sectors like finance with an unbreakable grip, (b) intertwine that grip with governments who want to circumvent civil liberties protections to suppress private action, and thus (c) become a de facto shadow government whose behavior - by virtue of being nominally private - isn't subject to constitutional protections or court oversight?

Because we know that not doing so leads to bad outcomes, and in some cases, to outright catastrophe.

A person who was not invested in subprime mortgages in 2006 had no skin in the game - yet the fact that others did invest in subprime mortgages created instabilities that threatened them. Virtually everyone agrees, in retrospect, that something should have been done then. But it wasn't, with precisely the justification you're articulating here. The problem is that that person did in fact have skin in the game, because the outcome had important ramifications for their life even if those ramifications were not in the form of direct financial losses.

Now, sure, your ability to buy furry porn is very different from sparking a global recession. But you're implicitly articulating a very strong claim here, that you cannot regulate economic activity to which you are not a party. That has a clear counterexample well within the memories of most people reading this thread.

------

I think we agree that private individuals should be able to purchase legal content from the people who produce it. Without action here, that will become either impossible or very difficult, to the point of having a major chilling effect. I think we also agree that businesses should generally have the right to conduct business as they see fit, both because it allows the exploration of new ideas and because market economics is a powerful force for increasing productivity.

To me, that says that there is a tension between two irreconcilable rights. On one side, we have the rights of businesses to act in their economic best interest (which is important!). On the other side, we have the rights of individuals to (actually and with reasonable effort) engage in lawful private microeconomic activity. And when you encounter such a tension, you need to consider:

- How important the rights are

- How much of one you get by sacrificing some of the other

In this case, I would consider the ability of individuals to conduct microeconomic activity more important than the ability of corporations to conduct what is effectively a PR campaign (since no one seems to be of the opinion that payment processors are actually taking a loss on people buying porn, they're just caving to political pressure). And I think the restriction of payment processors here is small compared to the potential restriction on private individuals. So to me, the trade-off has a clear winner.

If you disagree with this chain of reasoning, can you explain where?

"Small" is doing a lot of work there.

Large entrenched companies have leverage small businesses do not, in the same way that a large moon orbits in a way a test particle of infinitesimal mass does not. We already recognize this with respect to monopoly law: you lose your right to do certain things to your competitors precisely when you're large enough that you could reasonably suppress them.

That is essentially what we are talking about here: a duopoly that is actively suppressing competition. My understanding is that the big-two payment processors don't just refuse to process certain payments, they also refuse to work with banks who work with payment processors who will. Assuming that I am correct in that understanding (I might not be, this is not my area of expertise), that would prevent (or at least hinders) someone from just saying "there is a market need here" and forming their own payment processor to fill that need. To me, that seems like a problem for the exact same reasons that monopolies are a problem, and regulating against monopolies is not particularly controversial.

I think there's also a sampling bias here? ATLAS, the survey that discovered the comet, is specifically looking for potential Earth impactors. One assumes that would involve looking close to Earth's own orbital plane.

We know how fast it is moving and how far from the Sun it is. If its velocity is greater than the escape velocity of the Sun at its current distance, it can't have gained that velocity just from its orbit around the Sun (because by definition an object in an [elliptical] orbit is traveling more slowly than escape velocity).

It is possible for a Solar System comet to be perturbed by other effects (like a close passage with Jupiter) into an escape orbit. But in those cases, the speed above escape velocity is small, and the orbit barely escapes. 3I/ATLAS is moving much, much faster than that, too fast for within-the-solar-system effects to explain it. It must therefore be interstellar.

Hubble's slew rate (the rate at which it can change the broad direction its camera points) is about 6 degrees per minute [1], or about 1/10 of a degree per second (I refuse to use the incredibly cursed unit "minutes per second"). It tracks a fine object slower than that, but that gives a reasonable order-of-magnitude estimate.

At even 1 AU of distance, an angular velocity of 1/10 of a degree per second requires a linear speed of about 0.87c. Needless to say, 3I/ATLAS is not moving that fast - if it were, it would be outputting about 100 TW, mostly as heat, just from slamming into the interplanetary medium at relativistic speeds [2].

[1] https://www.pbs.org/deepspace/hubble/diagram.html#:~:text=Th...

[2] https://www.wolframalpha.com/input?i=%2840+*+mass+of+hydroge...

Assuming the bands are narrow, that should allow approximately true-color images, shouldn't it?

Human S cone channel = sum over bands of (intensity in that band) * (human S-cone sensitivity in that channel)

and similarly for M and L cone channels, which goes to the integral representing true color in the limit.

Are the bands too wide for this to work?