Traster

Woah woah woah woah woah. You've missed the obvious conclusion of what Adam Jackson CEO of Braintrust is saying. The obvious conclusion is that Adam Jackson is a liar. Oh yeah, you would think that this AI slop bucket at the front of our interview process would deter people - but the guy whose stock compensation depends on it working is very happy to lie that it doesn't.

Funny. The story here is that Youtube is asking every customer for legal id - presumably to enhance their adtech stack and target those customers more accurately. Obviously this isn't a great story, so they've screamed "won't somebody think of the children".

I do admire how much they dress it up though, they've developed tools to stop 17 year olds from repeatedly viewing body dismorphia videos that lead to annorexia. Sure, they could've tweaked the algorithm to stop paying people to create annorexia content, but then how would they monetize that 18 year old annorexic girl.

There's so much here, and not necessarily in a good way. The way this guy talks sounds a lot like those old effective alturist arguments that went along the lines of "Well if there's a 1% chance we can save a billion lives a thousand years in the future, that's actually better than saving 100 lives today". Ignoring the fact that "1%" wasn't an estimate that you could have any confidence in.

Sure, if Deepmind could save a few percentage points on their data centres that would be huge! Becuase you've taken a small number you have no basis for (a few percentage points) and timesed it by the largest number you can find! Hey Presto! Big number! But then surely the guys at Google are morons right - because they only bought 1 Deepmind, they should've been throwing hundreds of millions around willy nilly! At these savings they can't afford not to!

Secondly, it might be true that it's difficult for you to compete with these companies that are hiring in teams of researchers for hundreds of millions, but what you're also doing is handing employees hundreds of millions of dollars. What are they going to do with that money other than throw it into angel investing? You're literally sowing the most fertile ground for startups in history.

I think we should actually be viewing this blow up in compensation in the context of the hangover of ZIRP and COVID. ZIRP basically made money in silicon valley free, tech companies could hire anyone they wanted at almost any comp and as long as there was growth there were no discount factors so they could effectively make infinite time horizon bets. Then covid happened and helicopter money came in to keep the economy going and Tech hired like crazy massively bloating lots of companies. But as things returned to normal, it became obvious that hiring had just been spending, and the returns weren't there for it. I think it's going to become clear over the long term that the same is happening here, Tech has tonnes of money so they're going to spend it, but 3 years down the line someone is going to do the accounting and I would bet you we end up back in the same spot that we did with Tech hiring in Covid - a long and painful unwind as companies have to return to reality.

If you go to nvidia's jobs website today you'll find they're actively hiring FPGA developers for ASIC prototyping. Obviously they're not dumping their 5090 RTL straight into some 10 metre wide FPGA chip. First, they grab the largest FPGA you can get your hands on - the FPGA vendors tend to have a couple of comically expensive comically large SKUs for specifically this purpose. Then you pop a few of them onto a development board and partition your design across the cluster of FPGAs with some custom interconnect, orchestration and DFT stuff. FPGAs offer quite a compelling way of getting test mileage vs simuluation/emulation in software.

I think this has put into words a reason why I bounced off using AI this way, when I need something done I often have a rough idea of how I want it done, and how AI does it often doesn't match what I want, but because it's gone off and written a 2,000 lines of code it's suddenly more work for me to go through and say "Ok, so first off, strip all these comments out, you're doubling the file with trivial explanations of simple code. I don't want X to be abstracted this way, I want that...." etc. And then when I give it feedback 2,000 lines of code suddenly switch to 700 lines of completely different code and I can't keep up. And I don't want my codebase full of disjoint scripts that I don't really understand and all have weirdly different approaches to the problem. I want an AI that I have similar opinions to, which is obviously tough. It's like working with someone on their first day.

I don't know if it's giving the tools less self-confidence per se, but I think it's exposing more the design process. Like ideally you want your designer to go "Ok, I'm thinking of this approach, i'll probably have these sorts of functions or classes, this state will be owned here" and we can approve that first, rather than going straight from prompt -> implementation.

It's well known reddit has been eyeing up an IPO, and as a result it's pretty unsurprising they're searching for ways to juice their metrics in the build up. Tells a nice growth story so you can dump stock on retail and it's not really important if you do long term damage to the business - because it'll no longer be your business. Can't imagine that would be a great place to work unless they very strongly align your compensation with the IPO (commonly employees get the oppsite, a lock-up period after IPO to make sure all the juice is squeezed by the time they get to sell).

I think it's actually not - they mean the side face of the wheel, https://images-stag.jazelc.com/uploads/theautopian-m2en/cher...

They're talking about the wheels, not the tyres.

They that the face which you put the bolts through is flat. On older cars it was concave.

This is... dense. And I think in some ways it's just giving credence to things we should just discard. For example,

>In 2001, he uses this method to plot a graph and then predicts unbounded life expectancy by 2011

and then Dan cites this sentence:

> In the eighteenth century, we added a few days every year to human longevity; during the nineteenth century we added a couple of weeks each year; and now we’re adding almost a half a year every year.

Like, let's lay aside the predictions. Between 1980 and 2000 US life expectancy increased by 3 years. So let's just shelve the question of the predictions, this person is clearly making shit up. Whoever is making this prediction is simply not getting even the most basic facts correct.

In some ways I think the approach of this article is backwards. You don't apply retrospective rigor to old predictions, you define a rigorous process for future predictions, because there's no way someone is going to luck into being rigorous, and I think most of the futurists are even aiming to be rigorous. A lot of futurists are essentially sci-fi writers. What they're doing is fundamentally entertainment rather than rigorous scientific method. So I don't see why you would set out to judge them by a rigorous scientific method.

In some ways I see a parallel between this and effective alturism - EAs often think other forms of charity are misguided, but there's no point judging non-EA giving by EA standards because people just fundamentally don't agree with the core ideas of EA.

Yeah the thing that struck me about his Twitter thread was the absurd scare quotes around the terms he clearly disagreed with. It doesn’t come across as well intentioned at all. It’s like a get out of jail free card for literally any behaviour.

I think it would be naive to say Joe doesn’t have a specific goal when picking his guests.

This sounds like a project to almost do the exact opposite of what is required in the market. The real need in the market today is a way of auditing the process by which someone creates a carbon credit. You need to be able to trust that this credit you bought it from someone else doing something genuinely carbon negative vs just either claiming credits for something they were going to do anyway or claiming credits for not doing something they were going to do anyway.

The worst way of doing that is creating a market place of tokens that make these heterogenous carbon sinks into fungible assets.

If you look into the announcement there's also some really funny stuff going on:

> including its revolutionary collaboration with the Centrifuge protocol, which will unlock the debt markets for carbon project developers,

The business model of Centrifuge is... well... loan sharks.

> its collaboration with the Layer 1 blockchain Celo, to provide $10 million of GNT to offset Celo’s emissions.

Nice that we got to the self-dealing portion of the business so quickly. So Celo will buy $10m of GNT to offset carbon, and in return GNT will launch on Celo, thus driving Celo up to cover the cost of buying the carbon credits. Someone, somewhere is going to be the bag holder for these tokens.

They've reinvented the funicular.

That's the humour, it's not "bold and radical" but obviously true.

I'm finding it difficult to engage with the "Let's pretend USD is a ponzi scheme" pre-condition of some of this.

I also think that the premise that RAI is distinct from ETH is... tenuous. The problem that UST-TERRA had was that it was trivially the same, but that means what we're saying is that if RAI succeeds to any significant extent then it puts ETH in a situation where it may also death spiral. I feel like I'm agreeing with Vitalik there.

>Another extreme case worth examining is where RAI becomes the primary appliation on Ethereum. In this case, a reduction in expected future demand for RAI would crater the price of ETH. In the extreme case, a cascade of liquidations is possible, leading to a messy collapse of the system. But RAI is far more robust against this possibility than a Terra-style system.

I think the conclusion of the 2nd experimnet is "Don't buy RAI". It's just that if the return on RAI is positive, well, buy it but know it's going to explode. If it's negative well... You're earning a negative return well done.

In your scenario are you suggesting that HelloFresh start sending people refunds unprompted? That doesn't exactly sound likely, but sure, they could stick a "Disclaimer: If you delete your details it may make it more difficult for us to refund you if we suddenly lose our minds and start inexplicably sending refunds to people unprompted".

If it really keeps credit card details, that’s way more than just a tombstone.

I’m glad you mentioned this, yes the CTO quit 6 months later.

For the purposes of quantifying this we weighted one of those worth 95%.

Quite recently I quit a company just as they were bringing in OKRs. The reason they bought in OKRs was because they felt that the engineering organisation was failing to meet the needs of the company. So they put together a list of Objectives and key results. They basically said "This is what we have to do to be successful". There were a few small hitches. Half the Objectives were outside the scope of the engineering team. We could execute perfectly, but if our internal customer fucked up, our objective would be blown. They explained this was because it doesn't matter if we meet some technical objective if it turns out that wasn't necessary for the company to make money. The second small hitch was they were diametrically opposed objectives. We were to going to increase our release cadence 10x. We were also going to reduce production issues by 100%. That's right, our objective was 0 production issues whilst massively increasing our release cadence with 0 extra resources. The third issue is that they were unacheivable - we were supposed to deliver a 3 year project that would take 10 people in 1 year with 5 people.

It missed that the reason the engineering org failed to deliver was that the internal customer would change the requirements of 6 month projects roughly once a week.

It was basically an exercise in trying to pin blame on engineers for the failure of the company. It didn't bother me too much because I was quitting anyway.

OKRs don't fix dysfunctional organisations.

>A social media platform may not censor a user, a user’s expression, or a user’s ability to receive the expression of another person based on:

So not only are they at fault if they prevent your tweets from being seen, they're at fault if they prevent someone in Texas seeing your tweets. Which would prevent what you're suggesting.

>Matt Damon, who compared the advent of virtual money to the development of aviation and spaceflight in a critically panned but widely seen Crypto.com ad last year, did not respond to requests to weigh in.

Sorry but what is this? Does anyone watching Matt Damon in a superb owl advert think "Oh Matt Damon is personally in favor of this" and not "Matt Damon is being paid to read this script".

Does this author think that George Clooney just really fucking loves Nespresso?

They can create a Texas edition, but they can't make their non-Texas edition unavailable in Texas. So they'll still be subject to law suits based on their main edition.

In this case this isn't people voting their own manifest destiny. This is a Texas law that decides a California company must operate in Texas, and that whilst operating in Texas it can't decide it's own moderation policies, and that if it does choose to moderate, it can get sued repeatedly in Texas. Oh, and not just for stuff people in Texas post - a person in Texas can sue this california company for censoring something a california person has posted because the Texas user must be allowed to see what the california person posted.

So to be clear, under this law if a california person posts a nude photo and twitter censors it, every single user in Texas can sue twitter, repeatedly in every different court, under the claim that the california person is being censored for their viewpoint that nudity is ok.

It wasn't under-investment, it was lack of focus. They spent their time running after new growth opportunities at the cost of their core business. They wanted to do IoT (low power they're not particularly good at and the margins are awful), 5G (they had some success here) and autonomous driving (some success but still comparatively tiny). Meanwhile their core business started to fail.

>Notwithstanding the foregoing, the Equity Investor shall be permitted to issue Tweets about the Merger or the transactions contemplated hereby so long as such Tweets do not disparage the Company or any of its Representatives.

Seems like "do not disparage the Company" might be at risk with:

>20% fake/spam accounts, while 4 times what Twitter claims, could be much higher.

>My offer was based on Twitter’s SEC filings being accurate.

>Yesterday, Twitter’s CEO publicly refused to show proof of <5%.

>This deal cannot move forward until he does.

I think it's technically a little early to be getting buyers remorse... since he's not yet bought it.

This is kind of funny because the SEC is basically saying Nvidia held their share price artificially high by doing this, but Nvidia also bought back billions of their own shares, by this logic they were overpaying.

So my understanding is it works like this: You take some Python you throw it at OpenAI and you see if the resulting C++ compiles. Doesn't the fact that you need to check the result even compiles also imply you probably want to check if the result is functionally correct also?

When I worked for a big US company it was very clear- you open an office in Canada because it’s cheaper. So, probably not.