SloppyDrive
At the same time, we are reducing the proportion of outsourced work, because a local team with LLMs are dramatically more capable... There so many benefits to how you can organize a smaller high-skill team. I would put a 10-man team with the current tooling on par with a 10 high skill + 80 low cost heads even 5 years ago.
I can now translate tasks that can be easily split to low cost regions, and not have to deal with the cultural issues of India. I can also rapidly generate work of similar (and improving) quality... I have no use for a meat-proxy to a llm.
I expect a severe crash in the Indian tech sector, before a minute proportion of the workers can start to build it up again.
I, and people like me create wealth. Some of that wealth is pooled at an administrative level that makes sense across a town, state, or nation. Its still wealth I created; and I expect it to be used to benefit me, ideally proportionally to the tax taken.
Unemployment can be justified as crime prevention; Disability and healthcare systems limit individual worst-case outcomes (especially if due to chance). Super-commuter subsidies do what?
Currently I live in a city with an OK pt network; in the a high density apartment. I chose this because I can catch a train to work, go drinking locally, and I dislike driving.
If I could rely on a driverless car, i would happily live further out in the suburbs, as the driverless car removes the upsides of density more than anything else... And I think this is a common sentiment, driven mostly by housing costs.
And then you have the cost of a trip, of owning vs rideshare... If its my car I can choose the furnishing, pay for fuel or power however is most efficient for me (eg solar), not have to pay for cleaning, and store my stuff in the car.
The generous idea is that "subspecies" does not provide an anthropologist a useful lens to look at humanity, therefore we do not classify.
The alternative is that "subspecies" is too close to "race" for scientists, publishers, and funding bodies to touch, so its deliberately ignored.
It doesnt matter if you are actually running missions, if the scale is so small and wasteful that its not meaningfully comparable to the aspirational future missions.
One launch a year is not even close to what we can manage with our current technology, to the point where the scope is too small to be legitimately worth doing.
Its not solely a matter of energy; its about opportunity for learning. The current scale is too small to be worth doing at all.
If it was a program of something like >50 payloads over a decade, that gives enough opportunity for refinement, in cost, safety, and scale manufacture methods to actually see something new.
I do not believe I have even seen "good value" output.
What is there to respect?
Either these tools dont do much, and its business as usual with the cycle of attempts to offshore failing.
Or the tools produce massive quantities of mediocre work, the low-cost developer is now obsolete. Markets that relied on quantity to sustain the industry evaporate. The problems hiring overseas are only amplified by this, leading to increased overhead finding the skilled developers. Other issues such as timezones, large team coordination, political risks, and cultural mismatch remain, and smother a shrinking sector.
Or the tools work very well; and if they produce better quality than you can then you are obsolete. Development generally as a concept is unrecognizable.
There are many reasons why; but simply there is no surplus of super developers anywhere; you might find one anywhere, but they will not stay low cost for very long... And if you try to structure your business to find any number of employees overseas you quickly become overwhelmed by the averages and cultural practices of an area.
I fully believe its possible to find a single competent Brazilian or Indian for cheaper than in the west, but I dont think its possible to structure a company in a way that you can hire 20.
The historical value of a low cost region is get crap work for few dollars. "Just barely good enough". AI can do this by itself.
If you add AI to poor work, you get more extremely shit work for less dollars. If you add AI to a skilled worker you get a large volume of OK to poor work.
Basically my suspicion is as the tools improve they make low skill regions obsolete first.
We always find that small teams of locals can do much much more than a team with an unlimited number of low cost "developers". Not just because the competence of low cost devs is poor, but also the structure of how you work changes for the worse with a bigger team, for the worse with a distributed team, and for the worse with a skill-diverse team.
Thats before you get into the cultural flaws of favored destinations like India.
So we have been able to argue things like add one local + ai is better than about 20-100 Indians, depending on role and business structure needed to manage low-competence low-trust "developers". So we are planning to completely on-shore in the near future.
The bean counters are happy, and the quality of the work is improving.
Either give a solid set of requirements that let a dev assume things about a windows 11 system (good hardware security, in particular), or fuck off entirely.
The ability to form militias is so important, that everyone should have the right to bear arms, in order to enable this.
The idea is that it prevents the idea of a "special militia" having some selection criteria, so the government of the day cant make qualifying for its group a requirement to own guns.
So the optimistic valuation is based on: Global ride share killer + Large car manufacturer + power infrastructure + Robotics.
Somehow the valuation is as though TSLA will succeeded early enough to entrench itself; if robotaxi wide rollout happens in the next six months, i would be happy to say TSLA is worth more than its current valuation. If it cant then at best 30% of current valuation.
See how poorly the sword one moves compared to the others.
Optimus hand dexterity is interesting; and the handling capacity is targeting useful weights (10s of kilos).
Boston dynamics is the most interesting; targeting "fit human" manipulation, 50kg. Their adaptable movement and walking are the best ive seen demoed as well.
S and Y are not special enough to do anything for the brand, they dont qualify as halo products anymore. Probably still wouldnt be that interesting even if refreshed.
CT is still interesting, it looks different and has some tech inside that seems worthwhile to iterate on.
And unlike traditional brands, tesla has FSD, Optimus, and Musk to do enough to keep the brand itself healthy.
My guess would be they are deciding what they can learn by iterating the CT, and might decide to drop it in a year or two when the roadster takes the halo role.
They will keep trying to improve on volume for 3 and Y.
The key thing for me is that it only works when the LLM is used for tasks below the devs skill level; It can speed up somebody good, but it also makes the output of low-skill devs much harder to deal with. The issues are more subtle, the volume is greater, and there is no human reasoning chain to follow when debugging.
So you combine that with a company that has staff in low skill regions, and uses outsourcing, and while there might be some high skill teams that got a speed up, the org is structured in a way that its irrelevant.
Yes, there are some security threats, but solving them is more valuable than trying to design a car around true firewalls.
And with the way we are moving to centralized one system architectures, the device that does video processing can be the same soc that does smart infotainment.
Smart connectivity essentially comes "for free" if the manufacturer wants to hit 5 safety stars, so its not going away, and will come to ICE cars as they modernize the vehicle architectures.
These issues with reports and junk contributions come about because there is a huge payoff for pretending to be part of the community, but the benefit from actually contributing is generally less direct.
I dont think you can solve this with "friction", because the people you want to dissuade are more tolerant to these kinds of barriers than the ones you want invite in.