Well, that clears it up. I thought there was a possibility that it was changed, but I wasn't thinking of it as something that's supposed to be changed. I would argue that making our own data to describe tasks and using it on all clients independently can give better results than having MTA sync the peds' actions, because high level actions allow more accurate predictions than low level ones. Think "follow a particular element" versus "accelerate and steer right". The former can give acceptable results with much longer sync intervals because the latter will usually become obsolete quickly. This is in contrast to player sync, where primitive actions are the highest level actions we have, which I guess makes player sync a more complicated subject than ped sync.