Maybe they tweaked a parameter to make it more aggressive.
I mean, that is what he suggested they will do. What that looks like and when of course I don’t know.
Overall i don't expect any change to overall success/fail rate.
I expect an improvement (1 out of three of the failures, a missed 5-second gap, would theoretically be solved).
But we’ll see. They are such small sample sizes and so dependent on traffic it is hard to say how any given sample will do.
increased assertiveness probably means it could be be a little better sometimes, and maybe a little worse in other cases where it misjudges and jumps out in the middle of traffic when it shouldn't
Increasing crossing speed (it is inconsistent and they should push to a peak of 15-20mph or so, rather than 8mph (!!!) or 12mph) would only improve safety, from what we have seen in all recent clips. Obviously there may be exceptions in cases of u-turning traffic, etc.
This peak speed increase can be achieved with very little to zero impact on initial jerk, which impacts the safety and comfort.
I mean do you really expect 10.69 -> 10.69.1 in 9 days is going to be significantly improved over 10.12.2->10.69 development over 3+ months?
I mean, there were huge improvements in 10.69 vs. prior versions (at least 3x reduction in fail rate, perhaps more in busy conditions). It would require a much smaller improvement to get above 90% (less than 2x).
The biggest challenge I see for them is the inconsistency of stopping pose, and the root cause of that. But increased assertiveness should fix a lot of what we have seen. There have been an incredible number of perfectly safe gaps in all the videos we saw recently.
Remember, this is an easy, low complexity turn.
I would not have taken the bet if I thought there was no way they could fail, though!
