He has admitted time and time again, on twitter and in interviews, that FSD has been a series of "local maximas" which each time given him and his team a false sense of reaching the full self driving capability. So yes, when pushing the edge of tech, it's not easy to estimate timelines and he's admitted that. What more do you want? For him to call it quits or get into magic and pull rabbits out of a hat?

A single space telescope, the James Webb's, which is far less involved than FSD (runs in controlled environment, albeit harsh, doesn't interact with anything but well known objects and is produced once with little cost optimization, since it's not sold in mass volumes and need not attract customers) has been delayed time and time again and run over budget by multiple factors, amounting to billions of dollars. No one said along the way that the project leaders were liars who knew they couldn't build the telescope but just wanted to spend tax payers money.