bartwr

"non-profit nature of OpenAI would not allow for it."

How so? Seeking more efficiency and cost-effectiveness absolutely does not conflict with a non-profit mission.

"Conflict of interest" is not defined by a bad outcome or malice. It is defined as the potential of those, human nature, and various cognitive biases. "Conflict of interest" is disclosing anything that could lead to a biased decision or lack of transparency. Can a CEO of 2 companies be objective about a contract between the two and when claiming that his company no2 is better than the competition?

And in the tech-specific case: As much as junior engineers would love to believe in "superior" solutions, tech decisions are seldom clear-cut. There are many trade-offs: cost, efficiency, memory use, throughput, latency, ease of use, cost of switching, and many more. You always have a pile of pros and cons. Sometimes, one is strong enough, but most of the time, it feels almost like guessing/intuition. And then the conflict of interest becomes especially concerning.

This sounds like an insane conflict of interest.

CEO of a company (or worse, non-profit!) and a member of its board creates another, for-profit company (in partial secrecy/lack of transparency) that the non-profit would eventually pay a lot of money. This is almost a fraudulent level of siphoning non-profit money.

Btw, this is hilarious - regular employees have non-competes in their contracts (sometimes void/illegal, depending on the local jurisdiction) and breaching them is an immediately fireable offense (sometimes leading to more severe consequences). You work on a small thing on the side? Better be careful, ask your manager/HR, risk it getting taken over by the company (luckily, IIUC this part is mostly illegal now in all jurisdictions that "matter" for tech).

But sitting on multiple boards where you have much more room and possibilites for creating conflicts of interest and damaging the company? All fine and common!

This completely misses the point that I'm using it as a tool and I am ok if this tool is sometimes wrong and I need to double check every. As a tool it saves me insane amounts of time (much more that I spend on verifying), a time saving and automation revolution comparable to web search, mobile, or Wikipedia.

(Btw, if I need to do a code review and verify all work of an intern ir a junior programmer, it doesn't mean their work is useless - it's valuable and a net win)

You don't have a use for a tool like this? Fine, your call. But stop this patronizing arrogant attitude.

How's your sleep and emotional state when using it like this? With an elimination half-life of 15h, 2-4 days a week it won't leave your body. I personally hated it for this reason - poor sleep, tossing and turning even with a sleep aid -> feeling emotionally unstable and even less productive the following day. Plus unable to recover from any moderate sports activity.

Being cynical, conspiracy, or half joking (or maybe all of the above? :) ): this means that the big tech bets on the orange guy coming back.

Let's entertain this idea: They have much better behavioral and sentiment analysis tools than anyone else (this is why they are so much better at ad targeting! Their ads work and everyone in marketing says are worth this extra cost). And if they knew he will (obviously not certain, just with a high enough probability), wouldn't they try to play a bit safer for him to avoid an immediate retaliation?

I used to be a daily smoker for over 2y and a casual/social smoker for close to a decade. Since moving to US, for the last decade, I don't smoke as "nobody" smokes here (compared to Europe). I smoke now 1-2 cigarettes a year, mostly when visiting Europe and with smoker friends. (I never got addicted and never had any problem quiting cold turkey; I smoked because it was great social relationship builder - some of my best friends are people met on cigarette breaks).

Sometimes I use nicotine gums / patches as a dieting aid (effective appetite suppressant).

And I want to warn against this premise of short half life. This is not true in general. There are active metabolites like cotinine with elimination half life of 20h+, depending on your liver enzymes, gender, ethnicity.

If I chew a nicotine gum around noon, I have trouble sleeping the whole following night with my brain unable to calm down, mild jaw clenching etc.

So yes, moldafinil is a terrible stimulant due to too long half life, but so is nicotine (for many). I definitely don't recommend for that purpose. Maybe for extreme (crash) dieting, but this is unhealthy in general.

Sadly things take so long because democratic rule of law law "has to" work slowly (you would not want "innocent" businesses closed overnight or citizens arrested because of allegations). It's not great, obviously, and both false positives and negatives happen.

The issue is that people and companies exploit it, "better ask for forgiveness than permission", and after years they clean up evidence, their CEOs say they "don't remember", and even if they get fined - it's too late and will not save destroyed, monopolized market. All those gig economy "disruptors" are the worst, but probably all the biggest companies do this.

I wish that there could be a possibility of a retribution payment - a monopolistic payment destroyed an $xyz bln competitor? They need to pay this amount of money, maybe to some publicly owned alternative/open source competition if it's too late.

One aspect of SBF's downfall that is not discussed enough, IMO: amphetamines (even taken for medical reasons/ADHD) release a ton of dopamine, seek more release, and make you do some weird stuff and seek risky behaviors (and reinforce them through reward system).

I recall his interviews and annoyed tone at uncomfortable questions - made me laugh, just like people on speed.

And absolutely avoid selegiline for "productivity", especially with amphetamines, this will turn anyone into a pathological compulsive person (e.g. a gambler).

I know everyone looks for a perfect "process" - and hence all agiles, scrums, whatever, but I am a bit cynical (or naive? not sure which one haha) and don't think there is one, all are bs (and implemented in a toxic org stay toxic), and all that is needed is a great team. :)

Basically a mix of people with different attitudes, different seniority levels, working great together, being curious, exploratory, but also responsible and keeping deadlines when pushed. I worked on some teams like this - like Marc Levoy's HDR+ team at Google, it was amazing. We had unlimited freedom to explore, but the team had very senior, responsible, and accountable people and when a deadline was approaching (sometimes self-imposed) were becoming very disciplined.

But one thing I am sure - if your goal is something creative and delightful - never let the business side, VCs and other "money squeezers" decide your process, features, goals.

I don't have the latest iPhone, but mine shows a "macro" icon immediately when it switches to the wide lens? And I can either move back, or turn it off.

The evidence - people's statements and experiences.

I worked there for 3y and can give "on the record" statement that it was the most toxic company I ever worked for. 80h work weeks, management shouting at people and calling their work "shit", constant lies, firing people on the spot for minor disagreements and many many more.

The latter is the reason why we launched Night Sight. And a ton of people were pissed that it's slow and their subjects won't stay still for so long.

We were not conflating user groups, just that if one group is literally 95% of the market and all photos taken, then you optimize the default app and the default mode for them and make the experience super polished. The remaining 5% of power users can use dedicated apps...

Before switching to big tech I worked 8y in video games - and agree.

Many practices of gamedevs - like designers literally just exploring stuff without any plan for years before commiting to any feature - would make SV directors and PMs (not to mention investors) scream. :) This is a double edged sword. On the one hand, this aimless exploration is needed in pursuit of real fun (as opposed to compulsive loops of mobile games). On the other hand, this leads to crunch, games being buggy and always delayed, and how most fun games always had very troubled production and sometimes literally destroy the studios/teams.

More organized companies I worked at (Ubisoft) had less crunch but also less fun games (cookie cutter AC formula). The messy toxic crunch workplaces (CD Projekt Red) I was at produced amazingly fun cult games...

I think the author doesn't understand consumer products, especially cameras and what most people care about (taking photos!) and need.

Generally, people don't care if the phone uses expensive 5x lens or not. I don't really know who does. :) They are at some position and want to quickly take a picture - phone refusing to take a photo or taking an out of focus one would be terrible for them and cause outrage. Image Quality is somewhat important, but 1000x less important than capturing the right moment. If you miss the moment - it's gone forever.

I worked as a researcher at the Google HDR+ team, developing algorithms for Pixel 2/3/4/5/6 camera. And consistently we would get both user studies/bugs/feedback, as well as guidance from PMs that responsiveness/latency and "taking any usable photo at all as fast as possible" was more important than the Image Quality. I'll admit, we didn't always deliver and often compromised (doing in software what everyone else was doing in hardware back then is HARD), but this principle blocked some cool IQ improvements from launching - and I think was the right call, even if some of those were mine (like high quality multi frame super res in every photo).

I have not used dating apps myself (have a single partner for 15y+ and generally was always meeting romantic interests through friends), but ~half of my friends got their partners through Tinder 5-10y ago. They were very happy with it - though this was in Europe, not in a tech hub.

Two of my close friends who don't have partners and still use Tinder said that in the last few years, they became useless unless you are a spending whale (enshittification). Full of bots, full of people keeping to make new accounts to take advantage of boosts at the beginning of the profile, needing to spend money to get any matches after this start period. Basically bait-and-switch model and pay-to-win, but with romantic life and self esteem - sounds absolutely cruel.

This quote says nothing about product, but about continuing research.

Most researchers don't care about products. Many don't even care about any practicality (just the scientific pursuit). How do I know? Well, I happen to work as Research Scientist. I worked at Google 5y. :) For the majority of my colleagues, "having to" support their work after it was published was a nuisance and they wanted to work on new cool stuff instead.

I have a simple answer for you - researchers measure impact differently. If they measured it primarily by shipping products, they would work as engineers, not research scientists. As a researcher you realize impact by researching and publishing (internally or externally) and are ok with the idea that it's up to others to pick it up and turn into cool products.

Hopefully they work at the same company, but if they don't - it's not your problem to solve, but VPs and above.

But yes, it can be frustrating if you're junior and take your work personally and attribute its success to its (and even worse, your own) value. After a few years of career, you learn to look at it from a distance.

Google publishes amazing stuff, people are paid great salaries, they have a ton of fun doing it, work life balance is way better than at startups - what's not to like?

But we already have this dream. It's called the internet. And video games. And movies. :)

The question of "metaverse" is: 1. Do we need a head strapped device? Does it add enough to the experience to justify it? 2. Do we need those experiences be more connected and owned by one rent collecting platform?

My answer to both is a firm no, both make it strictly worse. But it's a personal answer, we'll see. ¯\_(ツ)_/¯

(Meta obviously wants it to be double "yes" to sell those head strapped devices and collect this rent. But this doesn't make it true)

You have clearly not programmed in any of those. :)

OpenCL was almost unusable due to how difficult it was to set up, plus limited support. Vulcan is terribly ugly and verbose and provides no benefits for pure compute workloads. Writing a "hello word" is a few thousands lines of code, debugging impossible.

Working with CUDA is pure pleasure - immediate setup, super easy CPU/GPU interop and code sharing, all the modern (at the time!) C++ features, super pleasant debugging with stepping, profilers, everything. :)

I totally wish others invested as much into tools as NVIDIA did - but they didn't and made the whole experience miserable. :( If your velocity is 1/10 because of terrible tooling, you need to expect some insane benefits to pick it.

Fascinating, so safety and cleaniness have nothing to do with the budget? Do you volunteer to clean trains and buses for free?

I keep telling my friends who get really into stock investment, read some stuff, invent strategies - and end up much worse (sometimes losing money) than me just dumping everything into a few almost random indices: 1. Greed (not as a pejorative or a judgmental term, just this itchy feeling that you want more than you have even if you gain and are not satisfied) is the fastest way to lose money, whether through a poor investment or being scammed. 2. When you decide to play a game of stock investment, who are you playing it against? Being better than average (just market returns) means being better than the average (not median!) player, who in this case is some institutional investor. Do they really think that with a few online courses they can be consistently better than people who do this stuff for a living?

I agree with both your points. The experience varies and depends a lot on the manager.

And about the cognitive dissonance - but to me, personally, it's not like there is one Google, but it's many companies with completely separate goals and ways of acting. And it's not just a metaphor; different orgs really do operate on sometimes opposing goals (like Android vs Pixel vs Research - I was in the middle of that, but won't elaborate publicly). I think long term it should be broken up by regulators, but that's a different and polarizing/political topic.

But back to my original point - a lot of people have amazing experiences there, so my question to the person posting it remains - how is everything about those jobs horrible? This brings me a huge cognitive dissonance, as it's so different from my and my colleagues experiences.

Why do they sound horrible? This is a serious, non provocative question. :)

I personally loved working for Google. Low workload, great work-life balance, amazing perks, mind blowing salary, potential to do really cool stuff that hundreds of millions will use, doing really cutting edge things that almost noone else can do, ridiculously good internal infrastructure and tooling so I can focus just on my tasks, kind and friendly colleagues, nice offices. Everything about my day job and things I worked on (at Google Research) was great.

I disapprove of many other things company does, but they were completely separate from my work.

I have many friends at Meta, company I disapprove way more - but their sentiment is exactly the same.

That last sentence is a total strawman. Unions are not there so you get everything you want. They are there to turn individual negotiations and protection into group and collective one. So it's not just a single junior programmer fighting for their rights, but potentially all company SWEs fighting for this junior programmer rights.

If Meta faced a complete strike similar to writers' strike, their workers could achieve similar goals like writers did.

Or not - but nobody promises it, that's silly.

The amount of people here saying it's ok or even a long lifespan (?!) is shocking.

At college, around 2008 I had electronic circuit labs where we used Soviet oscilloscopes dating back to the 70s. They worked perfectly well (we couldn't read the Cyryllic, but after 10min intro you understand the knobs). And why wouldn't they work?

Stockholm syndrome of some is unbelievable. And then when I say that I don't want any cloud-controlled devices, people tell me I'm paranoid...

If Google is to retire hardware like this, they should either open source the full stack, write new OSS version of it, or refund the customers plus be fined in tens of millions for e-waste. (While they green wash).

PS. I used Jamboards at Google and they were cool for meetings or job interviews.

I worked at Google and I didn't get that impression. At no training I was told to not try to overtake the market, at many meetings PMs and directors were strategizing how to overtake the market, but the trainings were all about avoiding words (especially in writing) that could be seen as trying to establish the market dominance.

I review 10-20 papers a year.

It's a ton of unpaid, volunteer work, if I want to be a high quality reviewer then it's at least a day (at least 3 thorough reads, taking notes, writing the review, reviewer discussions, post rebuttal, back-and-forth for journals). I am lucky and privileged that my employer counts this towards work time. Only 20% papers get accepted in my domain.

Now if I had to spend a week on replicating a paper - and this is CS/graphics, where it's easy and "free" - I'd never volunteer to being a reviewer.

You'd need professional "replicators", but who will pay for them? And who will be them - you need experts, and if you are an expert, you don't want to merely replicate others people work full time, instead of working on your own innovation.

What a terrible advice (sorry). Great comments are not about "what does this code do?". They are about "why does it do it?".

In some domains like scientific or high performance computing they are absolutely necessary to a) provide context to others b) make you not forget c) prevent unintended consequences and regressions.

You reorder some operations or do seemingly weird things for improved performance, numerical stability, or address some not easily testable bug? If you don't comment it, this knowledge will be lost and someone might obliterate it during a refactor.

As a user of a product, I do not want to be a community of random people and flooded with unproductive discussions.

I do not want to be refreshing mostly-synchronous chats and if missing a day, missing important information.

I want companies to respect my time and attention.