Anecdotal evidence is empirical evidence. It's just not very convincing because it doesn't come with any information about how common experiences like those are, or what confounding factors might be at play.
Now, in this case, confounding factors probably aren't a problem, but if 10x developers are one in a million or one in a billion, an anecdote about how someone, somewhere actually met one once isn't very helpful to the rest of us.
Validity of the observation is a potential confounding factor. There are clearly people who impress their coworkers to the extent that those coworkers believe the person is a 10x-average programmer. That's what this kind of report establishes. But is the judgment of those coworkers correct? Is the person actually 10x more productive in any kind of rigorous, verifiable sense? Generally these reports don't come with enough evidence to establish that.
I believe you meant "He's anecdotally wrong because I've worked with 10x developers."
To be empirically wrong, you'd have to contradict it with some empirical rather than anecdotal evidence.