pardoned_turkey

Focusing on the user is easy when you have little to lose. When you have a trillion-dollar business and 200,000 employees you're responsible for, a large part of your focus is not destroying that. And quite often, it's not easy to reconcile that with what your users might want.

As the other commenter mentioned, silos are not inherently bad. Indeed, in a large company, they're necessary to avoid dysfunction. You want stable groups of competent people who share priorities and lore, who own well-defined parts of the business, and who have the autonomy to set the strategy for their thing.

"Founder-led" companies are more dynamic mostly because they're smaller. Once they get to 100,000 employees, they will not be distinguishable from Google, Apple, or Microsoft.

Layoffs at Google didn't happen because they had to happen. They happened because the leadership was concerned that in the good years, the company accumulated way too much dead weight - pointless projects, underperforming employees that the managers never had to deal with because they could always hire more people, etc. It's an awful fix and only a temporary one, but unnecessary risk-taking can jeopardize a lot more than that.

For example, let's say you have an idea for replacing online ads with a better monetization system for the benefit of the user. How do you pitch that at Google? A misstep here could literally destroy the company. It's insanity, akin to Exxon selling off their fossil fuel operation to try their hand at making solar panels.

Regulatory and PR risks are similarly grave. For example, Google couldn't have pulled off something like TikTok without all kinds of regulators jumping at their throats right away. They had to wait for ByteDance to clear the way and then launched their own "also-ran" clone. It's the same story with ChatGPT: Google had the tech but not the freedom to let it loose.

All of this is rational. You can get away with a lot more when you're a scrappy startup and don't have much to lose. When you're a multi-trillion-dollar company, the math ain't the same.

I think it goes back to org structure ossification, but also keep in mind that in a sufficiently large company, every department is a thorn in someone else's backside. A world where the people you dislike regularly get the boot is also a world where you have to constantly justify your own existence, where you have aggressive stack ranks, and so forth.

It's a bit of a damned if you do, damned if you don't kind of a deal.

Ian's post is pretty incisive, although I've read so many of these over the past 15 years or so. And the prescription is always to go back in time.

I don't really think that's possible. When you're a newcomer, a disruptor, the whole point is to be different. You're bold, you have a clarity of purpose, you say things like "we're building a new kind of a company" or "the user comes first."

But once you achieve market dominance, your priorities have to shift. It's no longer "why wouldn't you try this" or "let's do the right thing." It's "why would you rock the boat and risk the nice thing we have?" It's not just about profit. Careers and incomes are at stake. People will get hurt.

Risk tolerance aside, your organizational structure ossifies too. When you have people who have been running processes or departments in a particular way for fifteen or twenty years, they have little desire to start over from scratch. And that's not necessarily a bad thing, because what's the alternative? A cutthroat corporate environment where you're never sure about the future of your job?

I think the only comedy here is that Google looked at these old-school companies like Microsoft or IBM and figured they can be different just because they "get it." And then, over time, they rediscovered the reasons why old companies always end up operating in a particular way.

It's probably a reflection of the fact that most engineers who joined OpenAI did so with the hope of becoming rich and working at what might end up being one of the most influential tech companies in the world. They probably figured the non-profit governance thing is just some weird accounting or PR thing.

I doubt they care about Sam Altman. But they didn't turn down offers from other tech giants to be stuck at a niche AI doomer research shop.

I'm not sure that's accurate. Most companies use "AI safety" as a shorthand for "brand safety", i.e., making sure that your LLM can't be trolled into praising Hitler or denying climate change. Basically, do what needs to be done to avoid bad press.

There is definitely that second and more abstract take - AI doomsday safety - that Altman liked to talk about to get favorable regulation passed. But it doesn't seem to be much of a real field. "Hard" AI doomers think that superhuman AI can't be reliably constrained and we should just stop. "Soft" doomers think we need to solve "alignment". But we don't really know what's going on inside an LLM, so we test for alignment by asking it what it thinks about Hitler. In other words, back to brand safety.

They dropped it right after ChatGPT came out. For a full year, they were luring top talent with promises of exorbitant equity, sold integrations left and right, played a role in writing the EO that makes it difficult for newcomers to compete with OpenAI, etc. Didn't we have a story about $10M comp packages at OpenAI just a week ago?

I can't imagine this was just Altman. I suspect that a large proportion of current OpenAI staff is in it for the money and couldn't care less about the non-profit goals of the pre-ChatGPT era. In fact, many probably oppose the idea because it stands in the way of a big payout down the line.

This coup was probably some sort of the last stand for the old-timers who still believe in the non-profit mission, but I doubt they can prevail in the long haul. Not without gutting the company, at the very least.

I frankly find the idea that we must shield adolescents from any form of work... just pretty bizarre.

There are some dangerous jobs out there, and I can perhaps accept the argument that most 15 year olds don't have the agency to sign up for that - although as a matter of practicality, having grown up poor, I know a lot of people who ended up better off working the kinds of jobs that journalists and the regulators now frown upon.

You can make abstract arguments about how in a post-scarcity utopia, no person should ever have to work. But we're not there now and you don't get there by preventing people from working before that vision of a 3D-printed, AI-based future can be realized.

For now, work is inevitable for most; so what's wrong about being paid a living wage for classifying content in front of your computer on your own schedule? As a teenager, I'd have loved that. And I was exposed to content far more vile than anything this gig could've thrown at me.

I don't think it's about understanding. Lawyers are pretty smart. But there's no upside to you as a corporate lawyer if you advocate for taking risks. Even if you think you're on solid legal footing, you're going to miscalculate sooner or later, or run into a hostile regulator. And then, it's on you.

Conversely, there's no real downside to being too conservative, especially if engineers and leadership are entirely deferential to you because they don't understand your field (or are too afraid to speak up.)

Although this is also somewhat true for security, privacy, and safety organizations, their remit tends to include "enabling business." A safety team that defaults to "you shouldn't be doing this" is not going to have much sway. A legal department might.

You're cherry-picking plastics here. The same can be said about just about any technology developed in the past 200 years, so what makes plastics special? We laugh at people who talk about the "precautionary principle" for vaccines or cling onto dubious research that purports to show some ill effects. Or, how about applying the precautionary principle to air travel - after all, do we really know the effects on your body of frequently changing timezones?

We've been using plastics for a long time. They improved the world in important ways. There's nothing we can detect in large-scale studies of industry workers or other exposed groups that would suggest they're dangerous. That's kind of it. Every story you see on HN about microplastics is hand-wavy and involves dubious assertions and hypotheticals, and the alternative is... what? That we go back to making all commodity items out of metal, with a more serious environmental impact, higher transportation costs, and so on?

There are wasteful uses of plastic which should be curtailed, e.g. for packaging fresh produce - but the concern here isn't microplastics, just pointless trash.

Interviewing takes time and effort, so it's best to focus your energy on promising leads. A strong endorsement from a person who worked with a candidate is a very promising signal. Without getting into the possible whys, "this stranger contacted me on LinkedIn instead of applying via the normal channels" usually isn't.

Of course, depends on how your pipeline looks like. But most tech job reqs get a lot of applications and passive leads.

It matters why you're making the referral. I've seen far too many referrals made just because a stranger or a social acquittance pinged an employee on LinkedIn. And the employee's thinking is "why not". After all, they're helping a person in need and there might be a referral bonus for them down the line.

I'm a recruiting coordinator or a hiring manager, there's nothing that signals to me that this is a good lead. I assume you're just doing this as a courtesy. I might take a cursory glance, but if the resume doesn't look particularly exciting, onto the pile it goes.

The "how" part matters too. If you're referring a person who you know is amazing at their job, say so, and be specific. "I spent five years working with them at Foobar Industries and they are easily in the top 1% of all engineers at that company" is likely to get attention. "I know this person and I think they may be a fit" isn't.

Sure, but I don't think it's about companies like Google doing their buddies a favor. It's more about not wanting to fight complex and uncertain legal battles with companies that have large and well-funded legal departments. Very few of us would be above that.

The vast majority of trains in the US are diesel-powered, although they're about 4x more efficient than trucks.

It's tempting to imagine they could all run on electricity, but it's probably hard, as many routes go through hundreds of miles of rugged nothingness. This is quite different from rail transport in Europe.

Wikipedia is... nuanced. Keep in mind that the entity doing the fundraising is the Wikimedia Foundation. They pay the hosting costs, but return nothing to the actual Wikipedians (editors, admins.) Instead, what's left is used to pay the salaries for hundreds of administrative employees, fund third-party charities, and so on. You can love Wikipedia but have misgivings about the Foundation.

The conversation here focuses on a distinction between "hard" real-time applications, where you probably don't want a general-purpose OS like Linux no matter what; and "soft" real-time applications like videoconferencing or audio playback, where you nothing terrible happens if you get a bit of stuttering or drop a couple of frames every now and then. The argument is that RT Linux would be a killer solution for that.

But you can do all these proposed "soft" use cases with embedded Linux today. It's not like low-latency software video or audio playback is not possible, or wasn't possible twenty years ago. You only run into problems on busy systems where non-preemptible I/O could regularly get in the way. That's seldom a concern in embedded environments.

I think there are compelling reasons for making the kernel fully-preemptible, giving people more control over scheduling, and so forth. But these reasons have relatively little to do with wanting Linux to supersede minimalistic realtime OSes or bare-metal code. It's just good hygiene that will result in an OS that, even in non-RT applications, behaves better under load.