Chris Lattner, Poached From Apple To Become Tesla's Top Software Executive, Quits After 6 Months (bizjournals.com) 140
Tesla said last night Chris Lattner, the vice president of Autopilot software, has left the company about six months after the electric car-maker hired him away from Apple. From a report: Lattner had led the software development team in charge of Autopilot. Tesla executive Jim Keller is now in charge of Autopilot hardware and software. The company announced it had also hired OpenAI research scientist Andrej Karpathy, who will serve as Tesla's new director of artificial intelligence and Tesla Vision. "Chris just wasn't the right fit for Tesla, and we've decided to make a change," the company told reporters in a statement. "We wish him the best." Lattner tweeted last night, "Turns out that Tesla isn't a good fit for me after all. I'm interested to hear about interesting roles for a seasoned engineering leader!" Lattner is a widely respected figure in the industry. He is the main author of LLVM as well as Apple's Swift programming language. We interviewed him earlier this year.
Explains Autopilot 2.0? (Score:1)
AP2.0 isn't even as good as 1.0 yet. The new hardware is all camera based instead of using a mixture of cameras and other sensors (radar/ultrasonic), and they expected it to be just as good with the right software and then move on to being fully antonymous.
At the moment it can't even stay in lane properly...
Re:Explains Autopilot 2.0? (Score:4, Funny)
"At the moment it can't even stay in lane properly..." OK in Boston, but might be a drawback elsewhere.
Re: Explains Autopilot 2.0? (Score:1)
Re: (Score:2)
Are you intentionally lying or just bad at putting across that you're pulling information out of your ass?
From Tesla's site:
"Eight surround cameras provide 360 degrees of visibility around the car at up to 250 meters of range. Twelve updated ultrasonic sensors complement this vision, allowing for detection of both hard and soft objects at nearly twice the distance of the prior system. A forward-facing radar with enhanced processing provides additional data about the world on a redundant wavelength that is a
Re: (Score:2)
Yes, but the AP2 system doesn't even use all the cameras yet. The radar is just for distance keeping cruise control, the AP doesn't use data from it for steering.
Engineer or Engineering Leader? (Score:5, Interesting)
He is the main author of LLVM
Way too many co-workers were forced or voluntarily tried the Engineer -> Engineering Leader route and turned out to hate it.
Code, unlike subordinates, does exactly what I tell it to do. If a mechanical design of mine fails it's because I screwed up not because my subordinate did.
Re:Engineer or Engineering Leader? (Score:5, Insightful)
Way too many co-workers were forced or voluntarily tried the Engineer -> Engineering Leader route and turned out to hate it.
Not all techs can lead, but some of them need to. Non-techs can't effectively lead technical teams. Some may claim that "leadership is leadership" but that is not true. You can't lead if you don't understand the issues.
Re: (Score:3)
Absolutely. The best managers I've had were engineers. The better ones were hands on engineers.
It still doesn't mean it's for everyone. Did Chris hate Tesla or did he hate being a Manager or being a manager at Tesla? Sometimes it just comes down to a clash of A-list personalities. He clearly knows his stuff and knows the direction he wants to go with a project. Jobs may have been hands off with the direction of LLVM while Musk would be more hands on with the direction he wanted Chris to do things.
Re: (Score:3)
He was a manager and a director at Apple, so its not like he was new to the challenge of engineering management.
Re: (Score:2)
Re: (Score:3)
Yeah, because techies hate challenging projects that advance the state of the art.
The regulation is part of the challenge. Prove it works on a private closed course. Convince the lawyers. The lawyers convince the politicians. The politicians change the laws and regulations.
But it all starts with making it work.
Re: (Score:2)
If you can't hop on the line and flip burgers, you don't know what flipping burgers correctly looks like, you can't gauge how much time and energy is needed to flip one, and your employees don't respect your ideas about how they should be flipping them.
Re: (Score:2)
I'm going to be a little pedantic, but I think it might be somewhat true that "leadership is leadership". However, if you don't know where you're going or how to get there, then it's likely you're going to lead people in the wrong direction.
Re: (Score:3)
Re: (Score:1)
Yea, too bad Steve Jobs as a non-tech person couldn't lead Apple....err...wait a minute...
Re: (Score:2)
Yea, too bad Steve Jobs as a non-tech person couldn't lead Apple....err...wait a minute...
Technical details were not Steve's strong point, but he knew how to code, and knew circuit board design. Before starting at Apple, he worked as a technician at Atari, where his boss said: "Steve was difficult but valuable. He was very often the smartest guy in the room, and he would let people know that."
Steve knew enough about engineering to know what was possible, and know when engineers were BSing him.
Re: (Score:2)
One of the better directors I've had knew nothing about the business, knew that he knew nothing about the business, and trusted the people who reported to him. It worked pretty well.
Re: (Score:2)
From the interview I heard (he was on the Accidental Tech Podcast a while back), he actually really liked his position as Lead, it's just that autopilot is actually a really interesting problem space and Tesla probably made him a pretty decent offer. (And, frankly, he's probably well off enough that the offer was more just an acknowledgement of his skill rather than any money he actually needed.)
Re: (Score:2)
He is the main author of LLVM
Way too many co-workers were forced or voluntarily tried the Engineer -> Engineering Leader route and turned out to hate it.
Code, unlike subordinates, does exactly what I tell it to do. If a mechanical design of mine fails it's because I screwed up not because my subordinate did.
You just haven't programmed your subordinates correctly. Go to a North Korean "teambuilding" camp for your next group outing and you will have them reprogrammed to follow your orders completely.
Re: (Score:3)
I have a different take. His previous development work existed purely in the digital domain. Even if he was successful running huge projects previously I think autonomous driving is a different beast that he was poorly prepared for.
For autonomous driving you need to integrate noisy sensor data from a chaotic real world environment and somehow use that data in a logic process that makes repeatably good decisions and fails gracefully. You need to understand the abilities and limits of the sensors, the rang
Re: Engineer or Engineering Leader? (Score:2)
Re: Engineer or Engineering Leader? (Score:1)
Re: (Score:2)
Why are people acting like Chris Lattner has never managed people before?
Re: (Score:2)
Re: (Score:2)
He was already "Senior Director, Developer Tools Department" at Apple. So it's not like he wasn't experienced in management. The "He's a coder not a manager" argument is false since he's spent most of the last half decade as a manager.
My money is on him being interested in AI and machine learning from a technical perspective and then once he got into it realized that working with machine learning is nothing like the clean, deterministic and pure code of a compiler.
I work with some software that is largely n
Quits? (Score:4, Insightful)
"Chris just wasn't the right fit for Tesla, and we've decided to make a change,"
He may have technically quit, but it kinda sounds like he didn't have a lot of choice in the matter. I also find it quite interesting that they talk about 'fit'. I doubt he is a high level asshole, given that post Jobs' Apple is not known for HR scandals, and in my experience personal issues have to be devastatingly bad before a company will let go of someone who knows their stuff. So maybe he didn't know his stuff (which bodes poorly for the Apple car), or he got off side with Musk (which bodes poorly for Tesla).
Anyway, I hope he negotiated one hell of a golden handshake. I don't imagine the Apple car project will be excited about having him back.
Re: (Score:3, Informative)
Well, the guy is known for compilers and languages. He was asked to make autonomous car software. I honestly don't see why anyone would assume because he could do one means he could do the other. Especially when compiler and language work is refining things which have existed for a very long time and autonomous car software is making something that's never been done before. They're two very different mindsets. I'm personally good at the first, but garbage at the second. But my ego is small enough that
Re: (Score:2)
Well, he would've been hired on as a tech lead, not just some random programmer. Understanding the problem space is probably hard, but not intractable for someone like him. He DOES have a PhD, after all. He might not be an expert in AI specifically, but he would've understood exactly how little he knew about the topic before he went in.
Re: (Score:2)
Well, he would've been hired on as a tech lead, not just some random programmer. Understanding the problem space is probably hard, but not intractable for someone like him. He DOES have a PhD, after all. He might not be an expert in AI specifically, but he would've understood exactly how little he knew about the topic before he went in.
AI is intractable. Doesn't matter how good an engineer someone is if the problem is unsolvable.
Re: (Score:2)
It's possible to solve it at least as well as a human, obviously.
Re:Quits? (Score:4, Funny)
Dad (a PhD) tells me that a doctorate is proof that someone can come at a problem and learn everything there is to know about it and then extend the world's knowledge, if only a little bit.
In my experience in industry, a doctorate is proof that a person could do the above, once, when they were young, not necessarily anymore. I've seen an awful lot of awful PhDs working, cluelessly, in fields outside their expertise that they treat as 'easy', while doing completely wrong. Some that can only be compared to 'know it all' teens.
Re: (Score:1)
Sure. Like pretty much any other achievement, some will repeat it many times over the course of a successful career, while others never will. Those in the latter group may not be suited for work outside the academic environment (particularly if they're easily distracted); they may rest on their laurels; they may grossly overestimate their capabilities; they may decide not to do the work.
And, of course, some are simply frauds. I personally know one or two PhDs who plagiarized a significant portion of their d
Re: (Score:2)
The military does lots of stuff in similar problem domains with little or no need for AI. There are lots of problems where Artificial Intelligence might produce great results, but commanding large, powerful, vehicles in a complex space with huge liability concerns isn't one that I personally would be all that comfortable with.
Dunning-Kruger effect (Score:2)
Well, the guy is known for compilers and languages. He was asked to make autonomous car software. I honestly don't see why anyone would assume because he could do one means he could do the other.
I don't mean to paint with too broad a brush but I've run into a fair number of software developers that are rather arrogant and presume that because they are talented at writing code that they are somehow domain experts in other fields as well. There are more than a few of them who have posted here on slashdot over the years. This doesn't describe the majority I think but it's hardly rare to see the Dunning-Kruger effect [wikipedia.org] among engineers.
Re: (Score:1)
You're kidding. They post naive slashdot comments on every story that's outside their field, all the while being certain that they have seen in five seconds what the "idiotic" researchers couldn't see in five years.
Re: Quits? (Score:3)
Why would anyone think he would be good at it? It's called the Peter Principle and it's truly evil. Fear it.
Re: Quits? (Score:2)
Re: (Score:2)
Re: (Score:2)
Anyway, I hope he negotiated one hell of a golden handshake.
Why? Is giving some rich guy a luxurious reward for failure really more important that leaving the money in a company that will spend it advancing technology?
Re: (Score:1)
Elon RUNS Tesla. He makes decisions and his teams have to make his vision happen (or they will find another job). Sometimes his decisions are based on his ego, like some of the shape and aesthetic demands he's imposed on his SpaceX engineers, and are not good engineering designs. But this doesn't matter. Elon's way or the highway. I would guess this is largely the source. Lattner had some vision of autonomy in his head and Elon shot that down. He probably felt he could not do his job the way he felt i
Re:Quits? (Score:5, Informative)
Or folks just disagreed strongly with each other and it doesn't really reflect poorly on anyone (at least until one or the other side is proven right or wrong).
In this level of doing things, compromise and moving forward may not be as feasible as it is in the lower levels. If you have leadership that really don't want to be on the same page, it will impact the quality of that leadership.
I have been in places where it has been very obvious that executives don't agree, and one is ostensibly yielding to the other, but it's clear that whether he meant it or not, his leadership was undermining the other because he didn't genuinely believe in the other's direction. Even as he tried to tow the party line and said mostly the right things, you couldn't help but to see his true thoughts bleed through and inspire decisions that did not work well in the context of the stated strategy.
Re: (Score:1)
"tow the party line" -> "toe the party line"
Re: Quits? (Score:1)
Spin from Elon (Score:2)
"They have hired people we've fired," Musk said. "We always jokingly call Apple the 'Tesla Graveyard.' If you don't make it at Tesla, you go work at Apple. I'm not kidding."
That's one way to spin it. It smacks a little of post hoc ergo propter hoc [wikipedia.org] though with a little bit of puffery [wikipedia.org] on top. Or it could be simply that they were a bad cultural fit at Tesla and Apple was a better fit. Just because someone doesn't fit at a given company doesn't mean they are inferior. I probably wouldn't be a great fit for Tesla either for a variety of reasons, none of which have anything to do with competence.
Re: (Score:2)
Re: (Score:2)
Turns out he got caught fucking the boss's wife!
Nope. The boss doesn't currently have a wife [wikipedia.org]. Maybe it was one of his ex-wives.
Re: (Score:2)
Turns out he got caught fucking the boss's wife!
Nope. The boss doesn't currently have a wife [wikipedia.org]. Maybe it was one of his ex-wives.
I can barely tell them apart. Even his current (or is that recent?) gal Amber Heard fits comfortably into the mold.
Haven't seen someone with such a strong type preference since John Derek (Ursula Andress / Linda Evans / Bo Derek)
seasoned roles (Score:1)
for a job hopper that will leave at the next glimpse of shiny
Re: (Score:2)
It's a farce! (Score:2)
Don't you see what's going on here people? Elon is making check knockoffs of people and enslaving the originals and eventually eating their brains! We need to save Chris from the evil clutches of the alien menace that is Elon Musk. WHO'S WITH ME?! ;)
Re: (Score:2)
after all, how much experience does he have with AI and autopilating software?
Autopilating software? That'll get the kinks out of your AI
Management wasting another good engineer? (Score:5, Insightful)
I've heard Tesla has a ruthless startup culture even though they're a huge company -- kind of the worst stereotypical SV startup taken to a new level because Elon Musk is so driven. If that's the reason he's out I'm not surprised. Coming from Apple where employees are pretty much pampered and living off the constant money flow from the App Store might be a pretty big shock.
But -- this guy is the "main author of LLVM". I've seen this one play out over and over, and have experienced it personally. Almost every company that isn't producing actual software treats their IT and development resources the same way the rest of the company regarding career path. Every individual worker in non-IT/non-engineering departments dreams of becoming the supervisor, then the manager, then the director and maybe a VP someday...mainly because most people aren't passionate about typical corporate jobs. The problem is that people management skills and engineering/work skills are completely orthogonal. No problem in the other departments -- who would want to be some random report analyst when they could be the boss of a bunch of report analysts and never have to see a report again? This is a bad fit for many engineers, scientists and IT people though, because most of us got into the job because we enjoy it.
Some companies are just starting to come around to the fact that not everyone is hard-wired for management and would rather just be doing more interesting and impactful technical work. That's how I've been able to structure my career (luckily.) When my current company figured out I was good at what I do, liked it and wanted to keep doing work like it, they gave me more responsibility on the technical side instead of a Kindergarten class of employees to manage. I'm hoping I can keep going in this way because I've done the whole department manager thing. I really tried liking it, but it's just not where my skills are best used. Being a senior engineer/architect type, teaching the newbies the ropes and figuring out our long term technical path is what I'm good at, and companies who figure this out with their smarter employees will benefit in the long run IMO.
Re: (Score:2)
He was a director at apple, he is not new to engineering management.
He was a people/technical director (Score:2)
From a long podcast interview he did around the time he left Apple, he has been mostly managing people for a while, along with some technical direction. But it was not like he simply had the title and was doing coding day to day.
Re: (Score:2)
Once you're working on something of sufficient size *you have to be managing other workers*. It's impossible for one person to do that much on their own.
If you are not managing anyone, that means what you are working on isn't that big.
Re: (Score:2, Flamebait)
Doesn't mean he was actually any good at it.
I've never worked for him or even met him, but I followed the Swift Evolution list after Swift was open sourced. When they first open sourced it, they announced a set of about seven priority goals for the next release (Swift 3). As the date for the beta arrived six months later, he sent out an email with a completely new replacement set of goals and then announced a great release that met the new goals. There was much congratulations and back slapping, but in rea
Re: (Score:2)
Now that's a novel use of the phrase "ABI stability". I guess you think Windows programs are statically linked with all required libraries? No? Well, Windows do have a stable ABI for both user programs and kernel drivers.
IOW I think the Anonymous gentleman that replied earlier wasn't completely wrong.
Re: (Score:2)
You're reading GP's comment backwards. His point is that Swift does not have a stable ABI, ergo programs must be statically linked.
Re: (Score:2)
Re: (Score:3)
Coming from Apple where employees are pretty much pampered and living off the constant money flow from the App Store might be a pretty big shock.
um... you mean, hardware sales?
Re: (Score:3)
App Store money is pittance - most of that goes to servers, credit card processors and other expenses. And online revenue at Apple (covering ALL of iTunes and iCloud) is tiny compared to even Mac sales. Apple's not getting rich off the App Store - 30% is not a big cut when most of the apps are free.
No, Apple is a driven workplace - maybe not ruthless, but one where
Re: (Score:2)
Apple: Highly Profitable
I'm pretty sure that profitability is not exclusive with presence of mediocre employees, especially in very large companies.
Perhaps he didn't code swiftly enough (Score:2)
Good news for Blackberry (Score:2)
He should give John Chen a call. Blackberry just got a 900 Million dollar windfall. I'm sure they are looking for good people for software development.
Main point (Score:1)
>the main author of LLVM as well as Apple's Swift programming language
The only point of Apple hiring him was to prevent LLVM going GPL.
Without LLVM, Apple would've been locked to GPLed realm of GNU C++ ecosystem, which meant no DRM and other goodies
Re: (Score:2)
That would be false. It was after hiring him that Apple was willing to re license LLVM as GPL to get it integrated into GCC.
The patch I'm working on is GPL licensed and copyright will be assigned to the FSF under the standard Apple copyright assignment.
https://gcc.gnu.org/ml/gcc/200... [gnu.org]
Re: Main point (Score:2)
No, they did not.
>copyright will be assigned to the FSF under the standard Apple copyright assignment
No talk of GPL here, they talked on a limited copyright, not a geniune change of license
Re: (Score:1)
My quote explicitly mentioned the GPL. Are you illiterate?
Re: (Score:2)
If they had assigned the copyright to the FSF, then the FSF could have distributed it under any licence they wanted.
Re: (Score:1)
The very next sentence after the one you quote says:
[quote]Initially, I intend to link the LLVM libraries in from the existing LLVM distribution, mainly to simplify my work. This code is licensed under a BSD-like license, and LLVM itself will not initially be assigned to the FSF.[/quote]
The patch that was being released under the GPL was the compatibility layer between LLVM and GCC, not LLVM itself.
I don't think LLVM would have gone GPL with or without Apple. The community developing it is happy with it bei
Re: (Score:1)
Sure, it might not have but Chris kept talking about doing a full copyright assignment if the community agreed.
If people are seriously in favor of LLVM being a long-term part of GCC, I personally believe that the LLVM community would agree to assign the copyright of LLVM itself to the FSF and we can work through these details.
Either way, the point of my post was that Apple's motives for hiring Chris had nothing to do with the ridiculous claims fubarr made. If Apple was all for preventing LLVM from going GPL why would their own employee in an official capacity be talking about a possible assignment of LLVM's copyright to the FSF. The copyright assignment would have allowed the FSF to freely relicense it as they pleased.
Re: Main point (Score:2)
Too young, too simple, sometimes naive.
The thing he was willing to GPL was not the LLVM, but a mere compatibility layer. A thing to keep exploiting GCC, while they make their own backend suck less. The whole text of the proposal sounds like a deal.
The legalese of that limited assignment basically meant that Apple can continue using GCC (because they discovered what a rotten tomato LLVM actually was), while being secure from all aspects of linking to GPL code. Same trick Nvidia did for 15 years with kernel m
Re: (Score:3)
That doesn't even make sense. FSF doesn't need copyright assignment to distribute LLVM under the terms of GPL. The BSD-like license allows for that already. They never promised to assign copyright of all of LLVM to the FSF. That wouldn't make sense because it would prevent anyone from distributing it under the less restrictive BSD-like license.
The patch to allow GCC to easily use LLVM as a backend needs to be GPL-licensed because it's a derivative of GCC which is itself GPL-licensed. Assigning copyrigh
Re: Main point (Score:2)
>FSF doesn't need copyright assignment to
>distribute LLVM under the terms of GPL
The whole thread was about the opposite case. LLVM wanted a deeper linking to GCC without having to relicense as GPL. Why you all guys want to derail this thread so much? Working for Apple, or their satellites?
Re: (Score:1)
Nowhere in Chris' post confirms that they weren't open to GPL licensing LLVM. In fact, it stated the opposite if the FSF had been willing to merge GCC and LLVM. The reason it never happened was the FSF refused not because Apple was against GPLing LLVM.
Re: (Score:1)
The thing he was willing to GPL was not the LLVM
Initially, yes. But as I quoted he was talking about a full copyright assignment which would have allowed the FSF to relicense it as they please.
They promised a complete asaignment, but people usually do not hold their promises, especially in business setting.
No, he said the LLVM would likely be willing to do a full assignment if the GCC people agreed to a merge.
I'm right, and you are not right.
Are you 2 years old?
Re: (Score:2)
You're a joke. High on your own immature ego you can't even understand simple things when documented, can't even write properly and have a lot of attitude with nothing to back it up.
Are you under 15?
Re: Main point (Score:2)
Yes, under a limited copyright assignment, _without_ license change
Re: (Score:1)
Keep shifting the goalposts.
Re: (Score:2)
To add to my other post.
Maybe that was true initially but Chris kept going on:
If people are seriously in favor of LLVM being a long-term part of GCC, I personally believe that the LLVM community would agree to assign the copyright of LLVM itself to the FSF and we can work through these details.
Headline and summary don't match (Score:1)
Re: (Score:1)
Tentacle Grape.
Re: (Score:1)
Giving or receiving ?
Re: (Score:2)