_jab

The details here remain unclear to me, and even this tweet is somewhat vague.

> I was given an offer that would explode same day. I had to forfeit all of my vested shares earned over my 3.5+ years at Windsurf. I was ultimately given a payout of only 1% of what my shares would have been worth at the time of the deal.

Was forfeiting the vested shares conditional on accepting the offer, or did he have no choice over the matter? Was the payout what he was offered as part of accepting the deal, or was that his consolation for not accepting it? The wording is genuinely unclear to me.

I literally see 3 interpretations here:

1. Offer was to forfeit shares in exchange for 1% payout, but OP rejected and still has shares

2. Offer was to forfeit shares in exchange for undisclosed payout, but OP rejected and got 1% payout instead and still has shares

3. He had to forfeit shares regardless of accepting offer, got 1% payout

(1) and (3) are both shitty offers from Google, but (2) is reasonable. Exploding offers are not uncommon in tech acquisitions. My guess is that (2) is what happened, since that's not in contradiction with prior reporting.

Three is already a lot of tools to be doing more or less the same thing. Not many people are going to have informed, experience-driven opinions on Cursor, Windsurf, and Claude Code, let alone the other tools you mention.

As boring an answer as it may be, mindshare matters. Newcomers face outsized inertia in winning that mindshare over from the major incumbents.

Let this be a learning lesson in judging these deals based on partial information. Kudos to the Google, Windsurf, and Cognition teams for keeping all of these deals under wraps until announcement (OpenAI could learn something...), but even so it's likely that we the public will never learn every detail of what transpired. I've seen a lot of harsh, misguided takes over the past few days, like that the Windsurf founders screwed over their employees, or that OpenAI reneged on the deal. In this case, this seems like a happy ending for all parties involved: congrats to the Windsurf team!

On the flipside, I’m pretty sure the investors got screwed.

Beavers are only endemic to North America and parts of Europe. So why does the rest of the world not overwhelmingly resemble the American southwest?

> How will you store my data and who can access it?

> The content covered by the court order is stored separately in a secure system. It’s protected under legal hold, meaning it can’t be accessed or used for purposes other than meeting legal obligations.

> Only a small, audited OpenAI legal and security team would be able to access this data as necessary to comply with our legal obligations.

So, by OpenAI's own admission, they are taking abundant and presumably effective steps to protect user privacy here? In the unlikely event that this data did somehow leak, I'd personally be blaming OpenAI, not the NYT.

Some of the other language in this post, like repeatedly calling the lawsuit "baseless", really makes this just read like an unconvincing attempt at a spin piece. Nothing to see here.

I once had a Go function that, unusually, was _expecting_ an error to be returned from an inner function, and so had to return an error (and do some other processing) if none was returned by the inner function, and return nil if the inner function did return an error.

In a nutshell, this meant I had to do `if err == nil { // return an error }` instead of `if err != nil { ... }`. It sounds simple when I break it down like this, but I accidentally wrote the latter instead of the former, and was apparently so desensitized to the latter construct that it actually took me ages to debug, because my brain simply did not consider that `if err != nil` was not supposed to be there.

I view this as an argument in favor of syntactic sugar for common expressions. Creating more distinction between `if err != nil` (extremely common) and `if err == nil` (quite uncommon) would have been a tangible benefit to me in this case.

I think the reason the goalposts keep shifting is really that these AI labs are illustrating Goodhart's law in action: https://en.wikipedia.org/wiki/Goodhart%27s_law

We're at a point where AI labs are fairly reliably able to create AI systems that can achieve good performance on given benchmarks, via targeted training efforts. You argue that it's not reasonable to expect that performance to generalize across other domains. But given that these same companies are trumpeting that AGI is around the corner, I think that's a fair expectation for us to have, and AI has disappointed in that regard.

Is fathering 28 offspring your idea of a better life?

A few thoughts:

1) I agree that the moat for these companies is thin. AFAICT, auto-complete, as opposed to agentic flows, is Cursor's primary feature that attracts users. This is probably harder than the author gives it credit for; figuring out what context to provide the model is a non-obvious problem - how do you tradeoff latency and model quality? Nonetheless, it's been implemented enough times that it's mostly just down to how good is the underlying model.

2) Speaking of models, I'm not sure it's been independently benchmarked yet, but GPT 4.1 on the surface looks like a reasonable contestant to back auto-complete functionality. Varun from Windsurf was even on the GPT 4.1 announcement livestream a few days ago, so it's clear Windsurf does intend to use them.

3) This is probably a stock deal, not a cash deal. Not sure why the author is so convinced this has to be $3B in cash paid for Windsurf. AFAIK that hasn't been reported anywhere.

4) If agentic flows do take off, data becomes a more meaningful moat. Having a platform like Cursor or Windsurf enables these companies to collect telemetry about _how_ users are coding that isn't possible just from looking at the repo, the finished product. It opens up interesting opportunities for RLHF and other methods to refine agentic flows. That could be part of the appeal here.

Binary search is the obvious example.

What it and your example have in common is that a significant constraint exists on the input. I can't imagine how a deterministic algorithm with unconstrained input can process that input in sublinear time, but I would love to learn otherwise.

My guess is that the car is looking for a place to pull over (perhaps because the rider pressed the "pull over" button. In this mode, the car hugs the right side of the road until it can find a spot to stop, which would normally work alright, except in this one particular spot where the "right side of the road" is only this tiny island.

So the car just circles around it indefinitely.

> One recent idea I've had is that many online subscription services should automatically pause if you stop using it.

Cool idea, but probably tough to enforce what “using it” means. I could see companies start sending newsletters to customers and calling that engagement

No serious operator would choose a provider with your implementation of that feature.

I wonder about OpenAI's moat. Thanks to advances in hardware and rapidly improving open-source ML frameworks, it's getting significantly easier and cheaper to replicate what they've built over time. That's not the case for most startups: it's not really any easier to build an Uber clone today than it was ten years ago.

OpenAI depends on spending vast amounts of money to stay a year or two ahead of the competition. I'm doubting whether that's a justified tradeoff.

I think your comment is emblematic of the very divide between Silicon Valley and the rest of the world that leads to frustration like what the artists here are expressing.

No one becomes an artist except because of passion for the work. It sure as hell isn’t for the money. That’s not so universally true in Silicon Valley, and I’m guessing you’re one of those people who views their job mostly as a means to fund the rest of their lifestyle.

Glad to see this development. The amount of FUD around AVs is too high, and allowing each individual municipality to set their own regulations for AVs would have been a ridiculous amount of red tape for these companies to deal with. Just to pull one particularly bad quote from this article:

> “I hope that, in the meantime, our communities do not suffer too much in terms of injuries and community damages due to the current regulatory gaps,” Cortese said in a statement.

What gaps? What injuries? What community damages? If someone can actually present statistics that these cars are more socially dangerous than an equivalent amount of Ubers and Lyfts, I would be very, very surprised.

When you strip away some of the more pretentious language, this article isn't really saying anything that extraordinary. If you just swap the phrase "information" with "content", i.e. "addiction to useless content", it becomes clear that the article is really just talking about the dangers of doomscrolling.

At best, it's a useful reminder that doomscrolling happens not only on Instagram and YouTube, but also on "respectable" sites like the NYT, Hacker News, and Wikipedia, and that does resonate with me. But I don't see anything else that the author is really adding to the large corpus of discussion on doomscrolling and Internet addiciton.

It shouldn't take a Vox article to ensure employees basic security over their compensation. The fact that this provision existed at all is exceptionally anti-employee.

OpenAI's terrible, horrible, no good, very bad month only continues to worsen.

It's pretty established now that they had some exceptionally anti-employee provisions in their exit policies to protect their fragile reputation. Sam Altman is bluntly a liar, and his credibility is gone.

Their stance as a pro-artist platform is a joke after the ScarJo fiasco, that clearly illustrates that creative consent was an afterthought. Litigation is assumed, and ScarJo is directly advocating for legislation to prevent this sort of fiasco in the future. Sam Altman's involvement is again evident from his trite "her" tweet.

And then they fired their "superalignment" safety team for good measure. As if to shred any last measure of doubt that this company is somehow more ethical than any other big tech company in their pursuit of AI.

Frankly, at this point, the board should fire Sam Altman again, this time for good. This is not the company that can, or should, usher humanity into the artificial intelligence era.

I'm increasingly of the opinion that publicly available models should be required to disclose the data they are trained on. Perhaps not necessarily the raw data itself, but at least a description of what the datasets are.

I get the sense that there would be more backlash against these models which would drive us more quickly towards a lasting resolution if people better understood how their data is being used.

The argument may be that having very large models that everyone uses is a bad idea, and that companies and even individuals should instead be empowered to create their own smaller models, trained on data they trust. This will only become more feasible as the technology progresses.

Sam and Greg don't appear to be getting their board seats back.

I'm pretty sure that these companies always try to give you at least one notification a day, so that you never log on to see nothing new.

Agreed that it's user-hostile, but it's presumably effective as a retention tactic.

There isn't even a number. It's been generally regarded as just flatly impossible using current technologies. Even a partial step towards reliable earthquake prediction would be a massive advancement.

See https://en.wikipedia.org/wiki/Earthquake_prediction#Difficul...

Comparing the entire company's revenue to the fine is a bit misleading. Arguably, the more relevant comparison is to consider how much marginal revenue the violating feature actually produced. Granted, probably no one besides Google knows that number, but it's surely orders of magnitude lower. If those numbers are comparable, this starts to look like a reasonable penalty.

Because it's all just types, there's nothing _to_ run. Instead, you'll want to hover your cursor over the `PGunX` types at the end of the file in the playground; this will cause the TS editor to compute that type and it should display a string corresponding to the board state.

“all conversations are encrypted … at rest” - why do conversations even need to _exist_ at rest? Seems sus to me

Small startups these days regularly train cutting edge LLMs. I don't think Figma is that far out-of-reach.

Care to elaborate? I know I'd been making the same assumption about American universities - Harvard, MIT, Berkeley, and Stanford do seem to represent a good share of the world's best universities.