Frustrated Musk Shakes Up Autopilot Team

I’m not sure about this…

1 Like

Same thing happened to the Starlink team. There are 60 satellites orbiting right now. :man_shrugging:

I think everybody knows that Musk’s timeline for FSD is incredibly “ambitious”. He doesn’t predict HW3 deviating from HW2 until Q4. But also thinks that it’ll be feature complete in Q4. :thinking:

It’s been 8 months of work on Advanced Summon since it was “nearly ready” and in that time AP has barely progressed or even possibly regressed. I feel like they pushed out NoAP as fast as they could, did a bug fix release and then redirected all resources to Advanced summon but instead the team has been tied up for the better part of the year on something that was “supposed” to be done in October shortly after NoAP.

1 Like

I think there is a fine line between being fired for not agreeing something is possible (within a timeframe) when it is actually possible vs. when it is not actually possible.

We will see which side is the case. I suspect Tesla misses their schedule by a year, and I think they will be worse off for having lost these people. I don’t think they are in a must-deliver scenario for full autonomy. Great autopilot alone (supervised lightly by humans) will radically change their trajectory.

1 Like

Yup this was my thought as well. Elon has been both right (rockets being able to land and be reused) and wrong (producing 10k vehicles per week in 2018) about what’s possible, and his timelines are obviously consistently late. It seems like laying off the team for not meeting his arbitrary timelines could be harsh. Admittedly, I’m in no position to really accurately judge. But it does concern me - Tesla bled talent as well during its Model 3 ramp, and I feel like that was not all necessary

Interesting read, but with stories like this I’m always left shrugging at the end. A journalist talks to two employees and gets facts and opinions from them. The article ends up being a bunch of fragments of information with little sense of the context, meaning, or importance. You get what feels like a semi-random handful of facts and employee opinions and try to assemble them into a narrative or model about what’s happening at the company. This is not a criticism of the reporting; it just seems to be the nature of trying to report on a secretive operation.

A few questions that, if we had answers, could help put things into perspective:

  • How does this compare to Sterling Anderson’s departure or Chris Lattner’s departure, or other previous departures? Is it a bigger shakeup, or about the same? What has been the turnover of employees in the past?

  • To what extent are employees leaving or being fired because they don’t agree with Elon’s goals and timeline, and to what extent is it a perceived problem with their technical skillset or their performance so far?

  • Do managers or higher-level engineers have any serious concerns about how development is going? What is the general mood of Autopilot teams? Are they feeling discouraged or agitated?

In February, the same reporter (Amir Efrati, who does the best investigative reporting on autonomy by far), said this:

I’m not sure what this means exactly. Just that Tesla won’t reach the goal of being “feature complete” by the end of 2019?

1 Like

I don’t understand it either. At Autonomy Day, AP software was demonstrated showing recognition of traffic lights and stop signs. It made 90 deg turns at intersections. These are clearly improvements that if not yet ready to release to the fleet, are indications of progress. Tesla is following an incremental, bottom-up development strategy for Autopilot. Progress will come in fits and starts, but as long as we are seeing semi-regular advances in capability, I don’t think there’s a need to be concerned. Goals are goals, not promises. Having led software teams myself, I understand how a development team can get off track or lose effectiveness depending on talent mix, personality conflicts, and burnout. A good manager will take action to correct the problem.

Managing a team against over-the-horizon goals presents a pretty terrible trade-off. If you don’t pursue any goals which the engineering team doesn’t buy into then you’re not going to be pushing the envelope. Engineers, generally, only sign up for things they have high confidence that they can execute. If you push forward without their buy in then they won’t feel personally committed to the goal and might underperform because of that. If you force them to commit then their work life can become hell.

As an engineer I was occasionally put into that latter situation by my management and our circumstances. At times I had to perform against objectives that didn’t seem possible with dreadful consequences for failure. I lost sleep, my health suffered from stress, and my personal relationships were damaged. Work became my whole life and it was not a pleasant life.
But at those times I also got much more accomplished in much less time than was the case when I was more comfortable. Sometimes I failed, but usually I managed to eke out a win in an unreasonable situation. I hated it and never felt like it was a good tradeoff for me. But it was unquestionably the right thing for the organization, for our customers, for the other employees, and for our stockholders. Me pushing the envelope allowed our startup to survive and eventually dominate our multi $B market, while 17 of our 19 competitors went out of business. I also benefitted eventually since the financial rewards allowed me to move on to other things that would not have been possible if the company had not succeeded.

When I look at Tesla through the lens of those experiences I see an organization that makes the hard choice even if is unfair to some of the staff. People frequently leave in those situations - right away or later when they burn out. Some people have to be pushed out if they get in the way of pushing the envelope and many people just cannot, and should not live that way. It’s brutal, but it’s also brutally effective. And if you want to accomplish the seemingly impossible it’s often necessary.

1 Like

I’ve worked with a manager like that as well and also went through a rough time. It was in finance, so completely different industry, but he had some Musk-like tendencies. I had the most miserable two years of my life then.

I’m in agreement that pushing people beyond their limits allows you to squeeze more out of people. But it does come at a cost. It leads to brain drain and constant training that’s required to get new people up to speed within the organization. There’s very little human consistency, which slows the development of processes.

While processes have their own negatives, the reason auto manufacturers focused on them so heavily was because it led to 1) better reliability and build quality, 2) increased production, 3) subsequent margin expansion (profits). Something worth considering as Tesla has struggled to generate a sustainable profit.

I’m not even necessarily saying that Tesla’s pressured environment is something that Elon should aim to move the culture away from. It has arguably led to one of the most innovative companies in one of the most cutthroat industries. But it has also come at a cost.

1 Like

Apparently Stuart Bowers has left Tesla:

1 Like