- Not to mention the partner who he made move to another country and then still wouldn’t tell anyone about. The more I think about this post the more insanely controlling the guy seems!
- In support of your point, after Hashicorp relicensed Terraform (thereby killing, or at least squeezing, a lot of Terraform consultancies who had collectively contributed a massive amount to Terraform and prompting the creation of OpenTofu), Oxide and Friends spent several episodes discussing what was reasonable for open source maintainers to do.
They had a great discussion with Kelsey Hightower about it[^1], and his answer (which I liked a lot) was basically just that maintainers’ only real obligation was to be transparent about governance. If you want to be a dictator and ignore bugs and contributions that aren’t personally compelling to you, that’s fine—but please just put that in the repo. That way, people who are trying to build a business on open source work and want customers, or to build trust with users for any other reason, can distinguish themselves from maintainers that don’t care (as is their right). Otherwise the reputation of Open Source as a whole suffers.
[^1]: The whole episode is good but here’s the part where Kelsey argues that people just need to be transparent: https://youtu.be/13ctYOu8TsA?si=cI8TwHX6tAaCwH6o&t=1320
- I understand that DeepMind is working on this too: https://deepmind.google/blog/genie-3-a-new-frontier-for-worl...
I wonder how their approaches and results compare?
- I love the author's argument, but this conclusion feels obvious to me—rationalizing emotional decisions is like the oldest human activity there is. Try asking somebody why they're whatever-religion or whatever-political-party that their parents, friends, or partner is/are. I further claim that much of the purpose of managers is to do this for individual contributors: give people a story that scaffolds a decision you hope they'll make: staying at the company and going along with whatever decision has just been announced.
I also don't think it's necessarily bad that people do this. The input to any decision a person makes includes their entire life experience up to that point[1]. How could an executive encode all that in some kind of pat logical explanation, and how could the also-human engineers at the company possibly digest such an explanation, and what could make it more compelling to them than their own life experiences? People need to get through life, though, so they need to make decisions. They can't fully rationalize every single one, but they want to feel at least OK about the decisions they're making, so they tell themselves and each other these incomplete little stories and get on with it. That managers scaffold this process with their own stories is a little manipulative, but how else could people cooperate enough to have companies? The whole process just seems intrinsically human to me.
The most important part of being an executive is understanding all of this and choosing to hire people who will ultimately make good strategic decisions for you. Don't hire a well-known Perl contributor as your CTO unless you like the idea of rewriting your product in Perl. If your company is dying because this has happened, my condolences but at least you're not alone.
Edit: I hadn't read this far when I wrote the comment but the author also literally says, "The moment you hire a Rust developer to evaluate languages, you’ve already chosen Rust." I guess I just disagree that it could work differently. Each of us possesses bounded knowledge and bounded rationality, and "which language is best", is probably too complicated for an individual to figure out, especially when you don't even know what the roadmap will be in a year—you'd have to build the company several times in several languages and compare the results (and the best engineers I've met do write code multiple times, but rarely more than twice IME). Each of us can only really know how we would solve the company's problems. Executives' job is to try and guess, and make decisions that are ideally optimal but at least internally consistent.
[^1] My favorite example of this, actually: even in the highly-rational field of scientific research, scientists have to decide whether a given body of evidence is dispositive of a particular theory, and the standards they apply likewise depend on who they are and what their life experience is. So, as Max Planck put it, science advances one funeral at a time.
- Given all the money spent on trying different educational models to achieve better outcomes, it's really gratifying to see a result suggesting that improvement is actually possible. I have a lot of teachers in my family and they tend to take the perspective that education is an engineering problem rather than a research problem. That is, any apparent progress is due to extra funding or filtering students or the like.
> "Those costs do not include anticipated savings from improved teacher morale and retention, a dynamic demonstrated in other data."
That seems like some kind of supportive evidence as well. Teachers should logically be happier when working inside a system optimized for teaching efficacy!
Personally: we put our child in a Montessori preschool because we liked its emphasis on self-directed learning (I kind of think all learning has to be self-directed on some level. Even a lecture requires you to listen to and think about the lecture, instead of something else). We later moved him to a Reggio Emilia program for non-pedagogical reasons (there were problems with the building that the Montessori school was in). They're definitely different—in Montessori, he mostly played on his own, and in Reggio we now see him in pairs and groups all the time. I have no idea which is better, but his teachers at the Reggio school seem to like it.
- Lot of finger pointing at the author in this thread. I googled her and found this article from nine months earlier (I think), complaining that her husband does jack shit around the house: https://www.theatlantic.com/ideas/archive/2025/01/fair-play-...
Seems like the more boring but more real story here is that this mom is really struggling to hold her career together and give her kids the care she clearly wishes she could because her husband is being lazy. To the haters in the thread: I think this article can be read as "avoiding UPFs is completely unrealistic for authors trying to establish themselves while functionally raising a baby and a toddler by themselves". Which, even as a perpetual proponent of the anti-UPF book "Ultra-Processed People," I kind of understand.
I get a lot of pride and satisfaction from being an involved dad. I do almost all of the cleaning, a fair amount of cooking, and probably 2/3 of the missing-work-because-no-childcare (and I try to put in a good amount of solo weekend time, to let my spouse catch up on work). A valuable life lesson I learned in Boy Scouts: if you're not doing about twice as much work as you think is fair, you're probably not doing enough.
- That’s interesting, thanks for posting an explanation.
There is a parallel strain of argument for the former:
- https://www.tinygnomes.com/qwiki.cgi?mode=previewSynd&uuid=B...
- https://medicalxpress.com/news/2024-06-brain-overgrowth-dict...
(I have my own theory, which is that a large brain increases the risk of ADHD rather than autism—a larger flow of thoughts and ideas requires more executive function to manage, and therefore more executive function is required to achieve the same attention span—but that ADHD is a kind of multiplier for autism, because social situations are more challenging to navigate if you can’t reliably stay focused on the social interaction you’re having.)
- Look, the people making this argument all come from a fairly aligned political unit. If you start hearing this perspective from a broad coalition of ideologically diverse groups, it would mean the thesis has become consensus. I agree with you that the guy's tone is not great, so here's a slightly different framing: https://archive.strongtowns.org/journal/2016/4/17/sprawl-is-... (strong towns' emphasis is on ending federal infrastructure loans that leave cities and suburbs with infrastructure they don't have the tax base to maintain over a generation. Which I think leads to similar conclusions but they're careful not to say that).
Their framing also highlights the important financial difference between children and sewers, though: the former get less expensive and more productive over time, and the latter do the opposite, which is how the problems happen.
- Honestly the biggest reason I became a pro-density advocate is because low-density communities are almost always tax-revenue-negative[1], given that the cost of things like roads, water and sewer infrastructure scale with land area. What NIMBYs often seem to want are the amenities of city living (like a sewer instead of a septic tank in your yard) without the people, and that just doesn’t work.
- It’s nice to know that other people out there are as grumpy as I am (or can be), but I don’t think it’s particularly healthy.
Something I think about sometimes: the abstractions in math can be awfully elegant. Learning about them is probably 60% of the fun of studying math. But they took hundreds of years (or thousands in some cases) to conceive and refine. All that time was spent writing weird—often awkward and often wrong—proofs and trying every other idea that doesn’t work as well, and throwing them out (and still getting stuck with some of them, like pi instead of tau).
Software is less than a hundred years old. Eventually software will be quite elegant and make a lot of sense. Until then, we’re trying every idea we can think of (yaml templates) to see what’s good. Once AI can invent a programming language and reproduce decades of software development in it to test its design ideas, we can skip to the end.
- There's this: https://bench.vortex.dev/, which links to https://github.com/vortex-data/vortex/tree/develop/bench-vor.... I haven't tried pulling the repo or anything but it seems like they might be runnable?
Of course I don't know what benchmarks or performance metrics they might have for the db layer, but it is something.
- > randomly talk about my issues with a random phd in the lab (great scientist with tens of thousands of citations) which quickly scans the data and notes that the voltage obtained by the system in the publication is literally impossible
I'm interested in the apparent contradiction between the "tens of thousands of citations" credential for your evidently competent labmate who caught the fraud, and the "only high impact numbers get citations, only citations get you a chance to progress in the academia pyramid" indictment of the evidently mediocre fraudster.
How much of the science that happens do you think is due to people like your labmate, who seems to have earned her citations organically with talent, and how much is due to the fraud? Are most citations still going to talented people?
My parents were both academics who built their careers in the 70s and 80s. I don't know that they had a perspective on fraud per se, but they did say that science used to be totally different—lots of money going into a smaller research community. But because each PI trains many PhDs, the research community has grown unsustainably and now funding is highly competitive. For the system to be sustainable, the vast majority of PhDs need to leave science, and there need to better exits for those people (or else there need to be many fewer PhD candidates and researchers need to stop relying on PhD candidates for labor). I wonder if the fraud is a consequence of this problem.
- I think your frustration is illustrative, actually: the reason mutable global state is bad is because in order to prove that a piece of code reading it is correct, you have to know what _the entire rest of the program_ is doing.
If, instead, you make the global variable immutable, make the state variable into a function argument, or even wrap the mutable global state in some kind of helper class, then you only need to know what the callers of certain functions are doing. The visibility of those functions can be limited. Caller behavior can be further constrained with assertions inside the function. All of these (can) make it easier to prove that the reader is correct.
I think most programmers already do this, actually; they just don't think of their decisions this way.
- Ah, got it, thanks!!
- I'm trying to work through the math here, and I don't understand why these two propositions are equivalent:
1) min_{x,y} |x-y|^2
2) = min_{x,y} dx ∈ A y ∈ B
What is 'd'? If d is much greater than |x-y|^2 at the actual (x, y) with minimal distance, and equal to |x-y|^2 at some other (x', y'), couldn't (2) yield a different, wrong solution? Is it implied that 'd' is a measure or something, such that it's somehow constrained or bounded to prevent this?d ≥ |x-y|^2 x ∈ A y ∈ B - > This creates a system where companies find it easier to fire good employees in bulk than to fire bad employees individually. The legal protections meant to prevent arbitrary termination end up enabling exactly that
I have some passing familiarity with how (California) law firms approach firing, which this article gave me cause to consider:
- they do fire unproductive people aggressively (law firms bill by the hour and attorneys are very expensive to employ, so it’s very obvious and financially meaningful to the business when someone isn’t contributing)
- when they fire someone, they’re very secretive about it. The person stays on the firm’s website for months, and if you call HR, they’ll say that the person still works there (they probably do, in some narrow technical sense). This makes it somewhat easier for the person to get a new job.
Also this a nit but the legal protections aren’t meant to prevent arbitrary termination, which is pretty explicitly legal. They’re meant to prevent discrimination.
Given all that, where are professors supposed to find and hire students who don't want to stay in academia themselves? I think a lot of these students wind up being aspiring immigrants, and I'm not surprised that a lot of them would also have a hard time finding a place for themselves after graduating and that many of them would leave. Also, the abstract seems to argue that that US still benefits greatly from this arrangement: "though the US share of global patent citations to graduates' science drops from 70% to 50% after migrating, it remains five times larger than the destination country share."