_ea1k

That depends on whether Gemini CLI counts. I've had generally bad experiences with it, but it is free for at least some usage.

YMMV, but I've had pretty good luck with just force closing it and launching again when getting errors like that. It doesn't necessarily mean the whole environment is corrupt, even though that is the recovery option that is presented.

It is very unreliable though. I hope Android 17 improves it, as other than the restart issues, I've generally found it to be very functional.

You know what's weird? This is a company that has been using the fleet api for quite a while now to monitor non-professional drivers using FSD on their daily commute, often while distracted doing other things. The latest versions even allow some phone usage.

And yet people are skeptical. I mean, they should be skeptical, given that the company is using this for marketing purposes. It doesn't make sense to just believe them.

But it is strange to see this healthy skepticism juxtaposed with the many unskeptical comments attached to recent Electrek articles with outlandish claims.

Sure, but we now have millions of miles of Tesla autopilot and FSD data in the hands of untrained and often semi-malicious end users as well. Out of that data, we've gotten flawed reports from Tesla claiming that it is dramatically safer, as well as independent renormalization that showed it to be at best about the same.

None of those millions of miles resulted in a smoking gun showing the cars to be even 2x worse.

And yet a badly written blog thinks they've shown them to be 3x worse with professional monitors? This is indeed an extraordinary claim.

I've always assumed that the answer to this would be no. However, I also always assumed that a huge space-based internet system would be both expensive and impractical for bandwidth and latency.

Starlink has largely defied those expectations thanks to their approach to optimize launch costs.

It is possible that I'm overlooking some similar fundamental advancement that would make this less impractical than it sounds. I'm still really skeptical.

Yeah, I remember being amazed at the immediate incremental compilation on save in Visual Age for Java many years ago. Today's neovim users have features that even the most advanced IDEs didn't have back then.

I think a lot of people in the industry forget just how much change has come from 30 years of incremental progress.

TBH, the comments here amaze me. The claim is that a human being paid to monitor a driver assistance feature is 3x more likely to crash than a human alone.

That needs extraordinary evidence. Instead the evidence is misleading guesses.

Good analysis. Just over a month ago, Electrek was posted here claiming that Teslas with humans were crashing 10x more than with humans alone.

That was based on a sample size of 9 crashes. In the month following that, they've added one more crash while also increasing the miles driven per month.

The headline could just as easily be about the dramatic decline in their crash rate! Or perhaps the data is just too small to analyze like this, and Electrek authors being their usual overly dramatic selves.

It is sad, but big sedans do not sell well and the X really needed to be replaced with something completely different. There are now several other 3 row EV SUVs competing with it, and even low volume ones (eg, R1S) outsell it easily.

Don't be surprised if something else takes its place as they do need something larger than Y and less expensive than X was.

It was estimated at >200k/year, but in reality is well under 50k/year. I'd say that is a failure compared to their guidance.

They are almost exclusively focused on autonomous cars, humanoid robots, and energy (batteries now, maybe more solar manufacturing later).

As much as I dislike it, I can't disagree with the business case here. They already have >300k monthly subscribers at about $100/month. That business will grow rapidly from here as well as the robotaxi business itself.

Within 2 years, this business will look radically different just because of these two changes.

S and X were a small fraction of Fremont already. The plant can do >500k units per year, but S/X were closer to 20k.

It sounds like this would be giving ~5% of the factory space to Optimus production, which seems reasonable.

A big part of it is insurance. Family coverage can easily cost $15-20k per year in the US. Avoiding the need to pay for this out of pocket drives a lot of people into less than optimal job dependence.

I'd say it is a success at being useful, but yeah it does seem like the code itself has been a bit of a mess.

I've used a version that had a bd stats and a bd status that both had almost the same content in slightly different formats. Later versions appear to have made them an alias for the same thing. I've also had a version where the daemon consistently failed to start and there were no symptoms other than every command taking 5 seconds. In general, the optimization with the daemon is a questionable choice. It doesn't really need to be _that_ fast.

And yet, even after all of that it still has managed to be useful and generally fairly reliable.

Is this shocking? Obviously including PHEVs helps a bit, but even outside of this it is exactly what should be happening. Their biggest sellers are SUVs, and at these price points, the EVs can be substantially than their ICE counterparts. For 2026, they probably won't even need the PHEVs to get there, since the Cayenne EV is the best EV that they've built so far.

I put it in a VM and had it build a really simple todo app for me the other day. It wasted so many tokens that I can't help but agree with you right now. And I could certainly have done the same thing with beads and opus in approximately the same amount of time.

However, the gas town one was almost completely hands off. I think my only interventions were due to how beta it was, so I had to help it work around its own bugs to keep from doing stupid things.

Other than that, it implemented exactly what I asked for in a workable fashion with effectively one prompt. It would have taken several prompts and course corrections to get the same result without it.

Other than the riskyness (it runs in dangerous permissions mode) and incredible cost inefficiency, I'd certainly use it.

To be fair, he's always been a little loopy. At least, I think this post of his was loopy: https://steve-yegge.blogspot.com/2007/06/that-old-marshmallo...

It was also one of my favorite posts of his and has aged incredibly well as my experience has grown.

+1 - I think this really needs to be emphasized more. Just like any relevant skills, soft skills are useful. It is great to build strength here.

It is never a sign of health when they become the main thing.

Cheap crossovers and compact SUVs tend to be >8s. In practice even that makes them sound faster than they are, as they require an aggressive launch and consistent high RPMs to get there.

The 6-7s EVs feel much faster than that!

Yes, pretty much. The torque curve also slopes down as rpm increases, so an EV with really weak low end torque will feel really bad on the highway.

Having said that, there are some that are fairly mediocre without being completely terrible. The FWD Equinox EV as well as the FWD EV9 are acceptable to some people, but also pretty slow cars.

Plenty of people that can afford them and that would likely benefit are not buying them because of preference.

In many cases, this is rational. Yeah, a used Model 3 with a great rate plan is probably worth it as a commuter, but what if you are an outlier? Even a low battery failure rate can be a risk, and extended warranties are expensive.

But also, some of it is irrational, as the FB comments effectively feed many false fear driven narratives.

Claude Code vehemently agrees.

EVs and cold climates are a bit of a mixed bag. On the one hand, if you want to road trip with snowy 30 mph headwinds, the charging times will be meaningfully worse. Not impossible, but definitely noticeable.

On the other hand, the traction control is fantastic and they tend to have the best preconditioning features so that you never have to get into a cold car for your commute.

For a lot of people, that second paragraph is far more important than that first for at least one of the cars in their household inventory.

I don't think it will change anything for Tesla, unless it lowers their costs through lowered tariffs somehow.

TBH, Tesla is in a tough position with their EVs in NA. They can't really build a cheap enough crossover/suv to compete directly with ICE RAV4, and virtually anything they do at >$50k would negatively impact their existing product sales. The base Model 3/Y are too expensive compared to ICE and have met tepid reviews because of their slightly odd mix of price and features.

So they've chosen instead to focus on autonomy and car hiring. I can't blame them for that. There's a huge potential for recurring revenue in that space and they've been positioning themselves to be in an excellent position to capture a lot of it over the next five years.

I'd expect that kind of thing to be the primary use case, tbh. Maybe even running whisper models?

If it could run whisper, it'd be a solid addition to a pi based home assistant setup.

Exactly, price is a huge problem. IIRC, the average selling price of F-150 is ~50k.

The extended range Lightning tended to be $60k and up. Sure, it had AWD, but lots of people didn't need that. The Cybertruck is even more expensive.

Both had huge preorders when they were announced at ~50k.

I wonder who gets to decide when something is "associated" with something else in a way that makes any and all uses of that thing a cancelable offense.

This mechanism sounds more dangerous than useful.

I thought this way for a while. I still do to a certain degree, but I'm starting to see the wisdom in hurrying off into the change.

The most advanced tooling today looks nothing like the tooling for writing software 3 years ago. We've got multi-agent orchestration with built in task and issue tracking, context management, and subagents now. There's a steep learning curve!

I'm not saying that everyone has to do it, as the tools are so nascent, but I think it is worthwhile to at least start understanding what the state of the art will look like in 12-24 months.

TBH, if everyone involved is invested in what is being said, you don't need to be in person and you don't need to shut laptops.

If that is the majority of your meetings, you are in a good place.

The mistake is to think the rules are what makes the meeting useful. Having the right audience, an agenda, and appropriate expectations for the outcomes are the useful things.

You don't understand. If I arrange everything into written words and send out an email with a link to the document, noone will read it.

Instead, I must invite 10 people to do other things while I talk on a zoom call! "Sorry, I was multitasking"