In fairness to the AI the game has to make some decisions about who to prioritize as you can't have 24 teams with 2 or 3 riders each in preferential position. IRL there probably aren't 24 teams with reasonable leaders but in the mangame it can happen so very small erosion in stats can lead to a very big difference in outcomes. And when the game also thinks it might need to protect sprinters this problem gets bigger.
But why does the game think it needs to protect sprinters when they have no shot at GC? IMO that is a bug, plain and simple. It would be easier to account for in positioning AI if that wasn't a feature.
Not saying this isn't a flaw, just that in races like this where there are hills but not that hard or other hybrid type races the AI gets confused about who to protect which complicates the problem.
I think JT's testing was inconclusive about the influence of having a sprinter on Skjelmose's outcomes but either way I think his marginal hill hurts him - if he was considered a true GC candidate he wouldn't have dropped back.
His sample sizes were small obviously, so in that sense, sure, anything would be inconclusive. But I think the fact that it happened each time in both in the real race, and in JT's initial testing, but then didn't happen in JT's tests after he removed Rodenberg, is pretty telling to me. Only one variable changed, and the entire AI posture changed with it.
In reading through the whole thread I may have missed his final conclusion. I am not arguing it isn't a problem, I think stage 7 of TA has a similar issue.
I don't disagree with your points, I think my argument is that it is more a new, admittedly bad, continuation of existing issues where marginal GC guys get screwed in some circumstances. The game cares a lot less about 10th+ in GC than the Mangame (and frankly it is hard to blame Cyanide for that, for standard gameplay it is probably not important). If we star seeing top 5 guys get this then I think that is a huge red flag. When top 10 to 20 guys have it happen, it is to me an extension of existing complications.
Ulrich Ulriksen wrote:
I did not notice this in the testing, but as JT said it is very hard to notice and evaluate for every situation in testing. Even this short way into the season we have run about 50 stages. That is a lot for a single individual in testing terms. I probably ran about the same number in testing but it was entirely two races (California and Qatar) as you have to run multiple test to be sure of patterns and also the PCM24 version was changing as we tested, so that made it more difficult. So to run every stage type and combination sufficiently to evaluate patterns isn't realistic.
I think it's worthy to quote this one more time, because it can't be stressed enough. Like I said on Discord as well, the same people doing so much for the game already took a bit of this on as well, but can't blame them for not being able to do one million scenario's and looking at every possible rider in every possible race. If you want to blame someone, might pick some guys who did know the game better because of playing it themselves or being in other sims, who decided not to say anything about stuff like this.
It certainly seems if your rider is outside of the top 15/10 favourites (depending on course, stage) they can indeed be ignored and strange decisions are made by the AI. Sadly, it might be the case 'more' than in previous version(s). But if you never really had a problem like this before, I can only say congratulations on having such a strong team, as it has certainly happened on many occasions to many of us in the past
But if you never really had a problem like this before, I can only say congratulations on having such a strong team, as it has certainly happened on many occasions to many of us in the past
It hasn't happened to me much, but I've probably been planning to avoid these situations for about 5 years now already. The game will always screw you over somewhere at some point, I just try to give it less chances if possible. But I might be in the minority with the way I approach it.
