... Did you even read my post? Where do you see anything indicating that I think Autobidder doesn't do its job well? I have no idea how well it does its job. What I said is that it isn't an example of particularly hard software, and that the hardest part (long term maintenance across divergent environments) is yet to come.
Yes, I did read your post. simply saying that it is not a hard software job is pretty solid evidence that you don't know much about grid management and dynamic grid supply algorithms. For a beginners view of Autobidder and grid management in a complex sourcing environment this QCon presentation gives a decent idea:
Tesla Virtual Power Plant
Anybody who's actually dealt with high precision, high accuracy near realtime multi-source, multi-use software systems knows that telecomm switching is a decent example of the 'easy' side of such systems. The hard ones have highly variable source, response times, transmission loss and highly variable transmission node capacities. Traditional highly refined grid management systems have worked on multi-minute response times because nothing could respond in smaller units of time.
The true transformational consequence of Autibidder within grid management was shown in the very first deployment, when nobody at all outside the Tesla team actually knew, so the Hornsdale Power Reserve (HPR) it astonished everyone:
Tesla big battery defies sceptics, sends industry bananas over performance - Energy Post
There is copious technical and non-technical information about this. In 2017 almost nobody thought wind power or solar power could become practical anytime less than two decades away. Within weeks of HPR a sudden failure proved the amazing capability of Autobidder with battery storage:
Tesla big battery outsmarts lumbering coal units after Loy Yang trips
That accomplishment has spawned the worldwide rush to install grid-level battery systems and the massive project to replace peaker plants as soon as possible with batteries plus Autobidder.
Several traditional grid management suppliers are devoting huge resources to trying to replicate Autobidder. TE is working flat out to increase supply and regularly replaces the world's largest battery storage plant with another one, Moss Landing ready to take that mantle within the next month or two.
So, if not "...particularly hard" then reusing first stage rockets is not really too difficult nor was using a jillion laptop battery cells to power a car. It all seems easy when somebody manages to do the impossible.
I don't want to give you a hard time, nor anybody else. I'm simply astonished by the degree to which many people minimize accomplishments they don't understand. Some such people also say that the Apple M1 chip is nothing special. It is very easy to assume that things which work so easily must be also easy to do.
That complacency is why only SpaceX figured our reusable rockets, and only a couple competitors, none major, have managed to replicate the accomplishment. That is probably why the 'simple' battery powered car hasn't really worked well for anybody other than Tesla, but they're now learning.
FSD is a harder job than grid management not so much because of software complexity but because the consequences of edge case failure are so dramatically bad. Methodologically the problems are closely analogous, error tolerance is the difference.
Frankly we all need to understand the practical effect on software design that results from a rise in fault intolerance and a fault-tolerant system. The complexity does not actually change, but the accumulation of edge cases and incorporation of those into as system becomes a wearisome arduous effort of data collection and labeling.
If we had a perfect world one person would have the genius of solving the mundane exemplified by Jeff Bezos coupled with the genius of solving the insoluble that is Elon Musk. In the meantime Tesla will be brilliant but with inconsistent customer service, very infrequently needed and irritating when it is.
So some fo us will ignore shortsellers because we know Tesla will continue with making impossibly large presses that allow eliminating dozens of parts each, making batteries more durable and making factories doing things never before seen, but largely invisible to typical users who think it's all mundane.
Securities analysts, even bulls, think Tesla is a car company.