This is the smart money take. They ran an EV calculation on the cost of half heartedly allowing self repair (but in a sneaky way in which surely favors them) with a reduced probability of legislation vs the probability of legislation being imposed on them and figured they can milk more money out of this scenario.
They will surely allow DIY in the most convoluted, inaccessible way humanly possible and try to keep this scenario as close as possible to the tractors being unrepairable.
roughly, imagine a decision tree where each end node has a set of probabilities and a utility points values per outcome probability. You multiply out the probability of each outcome with it's point value and pick the decision path which maximizes your probability weighted outcome. In most cases utility points equals how much money you make or lose but it doesn't have to be that.
So if one path has a 10% chance of -100000 points (ie something terrible happens) and a 90% chance of get 1000 points (something very good happens) this path has an expected value of (.1 * -100000) + (.9 * 1000) = -9100
whereas some other branch has a 100% chance of an outcome of 10 points. You take the branch where you get the guaranteed 10 points because even though the best case is on the other branch, the average case on that branch is worse than on the 10 point branch.
And what I'm saying is they decided that their highest average payoff as doing some compromised version of self allowed diy whereas they get a really bad payoff in the case where the diy legislation is written externally and actually forces them to allow reasonable diy
They will surely allow DIY in the most convoluted, inaccessible way humanly possible and try to keep this scenario as close as possible to the tractors being unrepairable.