> a self-driving car really only needs to be better than the worst humans to be an improvement

I think this is false because varying abilities of people will use the autopilot, not just the worst. The autopilot needs to be better than average for an overall improvement. I think it needs to be much better than average to be compelling enough for most people to adopt it. Most people consider themselves to be good drivers, even with sufficient evidence to the contrary.

> I think this is false because varying abilities of people will use the autopilot, not just the worst.

That depends entirely on how effective it is and what incentives/disincentives there are (i.e. subsidies, taxes, insurance), which will itself depend on how effective it is.

But what happens when the road is suddenly closed for an accident? Or new construction appears? What will database-based self-driving cars do then?

They would handle it the exact same way they handle any obstacle in front of them. Which is a lot easier than figuring out the speed limit, and the lane demarcations, and which lanes go where, and oh the stripes are covered by snow, and oh the stripes are worn away, and ...

No one ever said that they could get all of their info from a database. But there is a significant amount of info which they could get from a database, which they might not be able to get at all otherwise.