I get the impression you aren’t a software engineer. So to clarify: AI is not generalized enough to the point where there’d be much common expertise unless, say, this code review was very narrowly focused on computer vision. And even then this article specifically talks about code review, which, unless you have the context of the Twitter codebase isn’t going to be very meaningful. You might do this if you’re buying a teeny tiny startup with no real engineering culture - not (formerly) publicly traded company.

I’be been a software engineer for nearly 2 decades and have been involved in multiple technical due-diligence endeavors. At best, you’re just grokking the big-picture and looking for any major red flags. Getting involved in individual code reviews is not a useful exercise outside of understanding a team’s SDLC and various coding practices - for an org the size of Twitter all of that stuff should have been documented and shared during due diligence (which Musk waved) - injecting outside engineers into the code review process is just an expensive and sloppy way to uncover what could otherwise be gleaned in an easier fashion.

I do see one big red flag for Twitter: it now has leadership that doesn’t trust the engineering organization. For a technology company that can be fatal if not resolved quickly.

Both the vision systems at Tesla and the bot-detection at Twitter are classifiers. Both would have trained on large datasets. Both would have domain specific feature detection sitting below a more general algorithm. Both would have a similar basis for evaluation. Both have real-time constraints on the classification problem. An engineer (who is probably an ML specialist) familiar with one would not be starting from scratch in understanding and evaluating the other system.

I hear they're rumored to be working on a Cybertruck as well.

I hear they're rumored to be working on a self driving as well.

> Counter-point: Tesla is rumored to be working on a phone.

How is this a counter point? How exactly doing superficial review of code they've never seen of a social network will help them with developing a phone?

Who has a better notification / messaging system at scale, that isn’t already a phone maker? I suspect the Tesla phone team would be thrilled to review that code.

> Who has a better notification / messaging system at scale

1. Apple. Facebook. TikTok. It's a somewhat simple problem that at scale is often solved by simply spinning a PubSub channel per user because you no longer care.

2. To review that code you don't just grab people to spend an afternoon doing superficial reviews of code they've never seen.

That's absolutely insane. Tesla employees are paid by Tesla to do work for Tesla. Freelancing during the workday at Twitter, SpaceX, or any of the other company just because Musk is the owner or CEO is basically wage theft by Musk.

How would you feel if he had Tesla assembly line workers renovate his house?

> That's absolutely insane. Tesla employees are paid by Tesla to do work for Tesla.

Unless Twitter is paying handsomely for the consultation?

Ah yea I’m sure Musk ran it by HR, compliance and legal.. had all the paperwork squared away on both sides and is making sure time tracking is done carefully.

or he just winged it and grabbed some engineers and said “come with me”…

Which is more in character for him?

> Musk ran it by HR, compliance and legal

Right? Tesla hasn't had General Counsel since 2019, and is not looking for one.

Tesla definitely seems like the kind of place where they love hiring HR people who habitually push against powerful higher ups.

It is still ethically a no-go. Tesla isn't a consultation firm and they haven't done this type of work ever except for personal interests of the CEO. It doesn't take a particularly skilled lawyer to establish a conflict of interest there.

Happens all the time and the only reason anyone cares about this time is because Musk. If Twitter pays a FMV to Tesla for any resources, and as long as the Tesla board doesn’t care (which they won’t), this is a non-issue. Doesn’t matter what kind of “firm” Tesla is.

If the CEO of Ford used Ford resources to purchase a rental property for themselves, people would definitely be talking about it. I would even flip your claim and say the only reason people are dismissing this is because it's Musk.

Maybe so.

And... you approve of this? Or what? What exactly is your message here?

I don't approve of minimum wage, but I'm still not going to be surprised when people pay minimum wage.

Of course GP realizes this which is why they were able to recognize and characterize the pattern. Objecting to something isn’t failure to recognize its existence.

I’m chastising an alleged shareholder who pronounces this as good.

You are chastising me for pronouncing it bad?

This is literally a wint tweet lol 6/1/14 the wise man bowed his head solemnly and spoke: "theres actually zero difference between good & bad things. you imbecile. you f*%#% moron"

They wouldn’t mind. The employees have expertise in renovating. So makes sense.

What a weird timeline we are living in..

It's not about the employees liking the work or not. It's about whether or not Tesla is paying them to do work that does not benefit Tesla.

This is basic, basic, basic corporate law. It's not a "weird timeline" and it is in no way specific to the tech industry.

Managers, directors, CEOs, etc. cannot have the corporation do services for them for free because this deprives the non-management shareholders of value. Using their position as manager to assign employees to work on their house is in effect causing the corporation to provide them with renovation services for free. This self-dealing would be grounds for a derivative suit.

Also, it only takes one shareholder (in principle) to make this an issue. Even if you and most Tesla longs think Musk earned the right to do a little self-dealing here and there, it is an open invitation for legal action by SEC and other shareholders.

If you owned 100% of the company then all of that wouldn't be an issue, but it's still not something any legal department would OK.

The comment above you is wrong to call it wage theft--that would refer to workers not being paid the higher of minimum or promised wage for the time worked.

I feel like there is a lot of misinformation here and am compelled to comment, I’ve started and administered two corporations.

What you’re describing as self dealing isn’t illegal, thousands of corporations are run this way every day as long as taxes are paid correctly on the fair value of any transactions between the companies in question. It doesn’t even have to be profitable, Tesla could categorize the work they are doing for Twitter 100 different ways; marketing, R&D, team building, training, whatever.

As for minority shareholders, they have completely arbitrary rights depending on the bylaws, articles of incorporation, etc. If they are lucky, they might be able to periodically vote for board members given the right class of shares. Otherwise they have pretty much 0 input into how a company is run.

It's a field trip. Is taking the team out to lunch also "stealing from shareholders"?

Are you real my equating eating with your team the same as doing work for another org?

The mental gymnastics of Musk supporters are only matched by the Trump supporters.

You can only lose by trying to have a conversation with them.

> Managers, directors, CEOs, etc. cannot have the corporation do services for them for free

Sure, but do you know this to be the case here? There are a whole lot of assumptions flying out here that Elon just went and poached a bunch of engineers from Tesla to work on Twitter on Tesla’s dime. Pretty sure no one commenting her on HN as any idea of what the deal and detail actually is, all we apparently know for sure from this article is that one or more Twitter folk think that one or more Tesla folks are looking at their code. Beyond that…no details are really known.

Agreed I muddied things with the phrase wage theft. It’s theft from shareholders.

Point taken, but it's not wage theft so long as he's paying them. He's stealing from Tesla stockholders. Though I imagine the engineers he borrowed have existing priorities and deadlines, so the experience for them and their teams probably isn't great.

Correct. I used wrong term. It’s theft from Tesla shareholders.

Ya, no, it may not be popular but it isn't illegal and the impact on Tesla will be neglible.

It may be poor options or even poor management but it isn't theft.

What if the developers wanted to this excursion?

“Hey, take some vacation days to do some lucrative consultation work” is probably how it went.

Tesla and Twitter both have experience in AI in the sense that Exxon and Unilever both have experience in chemistry.

Hey now, it would be perfectly cromulent for Unilever and Exxon chemists see real value in a collaboration around, say, a higher-efficiency fracking solution.

The well water then can he reused to make some purple drink

The key is whether or not Twitter is compensating Tesla (and therefore its shareholders) at a fair market value for the time and resources used by Tesla employees for the benefit of Twitter (and therefore Musk as its owner). Not whether or not it makes sense to do so.

If its a dumb idea on Twitter/Musk's part but Tesla is fully compensated for their engineers' time, all is OK.

If its a great idea but Musk just assigned Tesla employees to do work that does not benefit Tesla and does benefit another company that he owns, that is an issue.

The scale of this is irrelevant. Absent fair compensation to Tesla, even one employee doing a day's work for Twitter is prohibited self-dealing. No one would bother filing a derivative suit over such a small amount, but SEC penalties could absolutely apply afaik.

There is no definition of "fully compensated"

Even if Twitter is paying much more than their salary costs, you still need to value the damage to the business by taking the crew away from the ship.

You can't just assume zero impact unless this was just a one-day workshop.

Unless Twitter is paying the Tesla for Tesla’s employees time to Tesla, whether you care or not, doesn’t matter. A part time CEO of a public company is using the public company’s resources on one of his private project. If I owned a business with someone and found out the other owner was using our company’s funds but told our employees to go mow his house’s lawn, I would consider that misappropriation, embezzlement even. We don’t know the details here but I would hope it’s above board.

Then why were they reviewing unrelated codebases? iOS, Android? This doesn't hold up

Are these the same Tesla AI experts that can't make FSD work?

Maybe they're the ones responsible for making the door handles work when the car is on fire. Or the ones that wrote the AI that runs over children or in front of trains.

The obvious solution is to have the tesla tweet at the child to get out of the way, or at the local fire department that there's a bbq'd customer who needs assistance.

Because Waymo has been able to?

What does that have to do with the price of tea in China?

It’s waymo able to

Yeah but you'd be asking for metrics, PR curves, labeled data, and things like this. Not code.

AI as practiced in self driving cars is dramatically different than what is practiced in bot detection.

Self driving has a well defined, static problem space, with inputs that don't change often (how often do they come out with a new street sign?).

Twitter is combatting distributed adversaries who are constantly adjusting their approach in evading detection.

They are certainly different domains, but can you justify the claim "Self driving has a well defined, static problem space"?

One of the things that makes safety critical applications like self driving so hard is they have such an abundance of low probability, high severity cases that it is very difficult to define/test them all.

I think it's static inasmuch as it isn't an inherently adversarial problem. The world isn't bent on thwarting self-driving cars. Bot authors are bent on subverting detection.

> The world isn't bent on thwarting self-driving cars.

Oh, that'll come.

That’s fair, but I don’t think it explains the “well defined” part of the claim.

They problem is so ill defined that it sometimes has to be modeled as random events instead of a deterministic physical process.

What I mean to convey by "well defined" is that even though the problem space of self driving cars is enormous, the success criteria of teaching a car to self drive is probably going to look pretty similar in 10 years to what it looks like now.

Bots on the other hand changes constantly - what is the definition of an abusive bot? What is the definition of spam? The adversaries on Twitter adapt not only their tactics but also their goals.

This has a material impact just no enough for folks to understand or complain about. Can't do that with fanboys around.

Not all it has a stellar reputation

In vision...