I believe the point he's making (which I don't nessecarily agree with) is that he's right about the changes in direction, but he tries to estimate the result in the context of what currently exists.

So for example, he got it right that mobile games would be big, but they were big on iOS which didn't exist at the time he tried to make use of it.

Nd he was right that relational databases would decline. At at the time not relational meant XML. He couldn't predict the NoSQL of today, because that just didn't exist.

Personally, I think that argument is just retroactively applying meanings that weren't there, however.