by NovaVeles » Sat 13 Feb 2021, 19:39:32
$this->bbcode_second_pass_quote('AdamB', '
')Well, he is talking about processes that have happened, and obviously in the past, they happened without regard to restrictions people thought in the day. The claimed peak oil in 1919 in the US, it certainly didn't stop cars from advancing beyond his 1913 example.
My beef with him tends to be his timing estimates, and banking on autonomous driving which doesn't appear to be popping up with the rapidity he assumes, and is needed, to make his overarching idea work.
Oh yeah his timing will be well off.
The Peak oil estimates in the 1910's was always going to miss the mark simply due to the lack of information, just like how Hubbard did with his, although he got much closer. Predictions of the future will be guide lines at best. This also leads into the issue that the 20th century was an era of monstrous abundance and growth per captia, this was the paradigm that all these theories and models were produced in - we don't have that same luxury this time around.
Which kind of partially works against me with the next argument.
When it comes to automation, a lot of folks in the field estimate it will be about another 25-30 years IF it is even possible to get to full autonomous. The problem is that, again, the economic model was applied to a technical field. Initial progress in the field did not continue at the initial pace as assumed. Low hanging fruit was gained via the use of machine learning but it is rapidly slowing in terms of gains. Essentially, since 1998, the year-over-year rate of performance improvement computer performance increase has been decreasing. We are now at the point were a 15% year over year increase is considered a big achievement.
This also comes back to the issue of Moores law. We are nearing the upper limits of what can be achieved with Silicon based computing, progress has slowed dramatically in the last decade and we are now bashing our heads up against what has been called 'Boltzmanns tyranny'. We can make slightly faster chips but it will decrease the overall life time and accuracy of these chips. Essentially quantum tunnelling of electrons means we can only go so small without introducing errors. The latest 5nm nodes being developed buts individual components at 13 silicon atoms wide. That is not much wiggle room for improvements - and this is the brain that will have to power full autonomous systems.
And don't get me started on quantum computers, they are not even in the same realm of what we are using today.
$this->bbcode_second_pass_quote('AdamB', '
')He lays out how these things follow an S-curve quite well. Did you catch that part?
Sorry it has been a few years since I saw the talk, I did not remember that part. S-curves are funny things, you cannot tell the pivot point until well after it has passed.
My own personal theory is that we are going into an inverted bell curve. Prices come down, platau around the 2020's and then increase as we have to use ever more costly materials to keep up demand.
')I already have. Two EVs in the garage. And his claim of lack of moving parts, lack of maintenance, etc etc, were spot on. I've got no interest in the autonomous driving part yet, my most recent EV is a first step into only battery land, and I want nothing to do with the autonomous part. Yet. Let others be the first adopters, I'll buy used and on the cheap after the bugs are worked out.