I don't know what the outcome will be, but I do think this is a worthy discussion as FSD success is effectively an option that will multiply the ultimate value of Tesla.There are some things that only require logistics to happen; some things that require legislative changes to happen; some things that the science is proven, but require some further engineering to happen.
The Tesla cars to date did not require any advancements in science, or changes to laws to happen. There were some new technologies that were developed, but everything was done from existing science. The key question about Tesla's success was whether a start up car company was going to be able to navigate the extremely treacherous terrain of getting new cars from the drawing board to mass production in an environment with a lot of competitors already in the market. It was an incredibly complex logistics challenge, but it hasn't required any cutting edge science.
Landing rockets also proved to be a difficult challenge for SpaceX, but Elon worked out the entire thing on his return flight from Moscow when he and some other early SpaceX people went there to buy Russian rockets.
Full self driving vehicles has required several revisions of the hardware just to get to the starting gate. Elon thought that the cars built in late 2016 had all the hardware necessary to do FSD, but later figured out that they had vastly underestimated the computing power needed. Tesla brought in the top processor designer in the world (his designs are at the core of both the current AMD Ryzen and the Intel i series processors). He redesigned the processor to run FSD from the ground up to meet the new requirements.
A few years after introduction of that processor, FSD still hasn't happened because the software task is staggeringly complex. FSD is probably over 90% there, but the last edge cases are probably driving the programmers nuts.
I haven't been involved in this specific area, but I have a degree in Electronic Engineering and have worked most of my career in embedded systems and driving hardware with software. I was in the middle of the electronics testing for the Boeing 777 back in the early 90s. My job was on the side of creating the simulators to test the hardware, but I worked with people who were designing the stuff to go on the planes.
FSD is one of the most complex software problems ever attempted. Definitely the most complex ever to go into a consumer product. And the most complex intended to be used in the most congested places of our modern world.
Regulators are going to allow the tech to advance, but laws will need to change before level 4 FSD. Most jurisdictions now make it illegal to do anything that distracts you from paying full attention to the road. These laws became much more strict in recent years to curb people from texting while driving. At minimum those laws will have to change to some degree to allow the driver to be able to do something else while the vehicle drives itself. Otherwise, there will be zero benefit to FSD over current AP most of the time and few people will use it. The only time I think of using AP is when stuck in traffic or on long stretches of highway when I want to give my arms a break. The car is too much fun to drive otherwise.
As far as regulations go, regulator tend to be fairly conservative in their approach and are slow to approve new technology. Many car companies including Tesla have been pushing to eliminate external rear view mirrors for years. The Model X prototype had cameras for mirrors. But regulators have only approved electronic mirrors in a few places. And that's more straightforward than FSD.
The things Tesla and SpaceX has achieved: the Gigafactories, the cars in production, and reusable spaceflight are all things that match Elon's original visions for those achievements. It took longer to get to those goals than he predicted, but there were no major detours that required scrapping what they had done and completely starting over.
With FSD the problem has proven to be far more complex than Elon originally predicted and it has required a complete redesign of the processing hardware and a complete rewrite of the software and they still haven't released it.
Elon is a Physicist by training and he is good at working through how physical things like a car, rocket, or factory is going to work with his basic principles approach, but software is a different animal. In areas where Physics is not the dominant science underlying the thing, he has been rather poor at predicting the future. He was very wrong about COVID and may have done more harm than good with his Tweets. The path to FSD has also been badly delayed.
He's missed deadline predictions before, so it's possible FSD is another case of this, but it seems different to me. Even if Tesla gets FSD working, there is still the regulatory hurdles. Even Elon admitted several years ago that before FSD can get done regulators will need to change some laws.
We're way off topic here and I've had my say on this. Time will tell on this issue. I believe my initial comment was that I wasn't sure FSD was ever going to happen. A true skeptic is someone who has doubts, not someone who nay says everything. I'm the former.
Elon Musk is a brilliant industrialist, but that doesn't make him a genius at everything. He has misses as well as successes. It's possible he may pull off FSD, but it's also possible that the problem is too difficult to get to level 5 reliably either technologically or legally. There are a lot of unknowns.
My unsophisticated opinion is that FSD will happen relatively quickly (next 5 years) primarily based on 2 things. The rate of improvement seen from watching FSD beta videos, and the exponential nature of improvement described in the wait but why article (although I note the irony of linking to an article that is already 6 years old).