You have a very good point, and I slightly disagree with that statement in the blog post. But I think the core point he was trying to make has some validity. Perhaps it could be more accurately (if less succinctly) said:

"The specific technologies you learn today will either fade into obscurity or else evolve dramatically. To remain at the high end, you must continue to either follow those evolutions, learn the newer technologies, or preferably both."

For instance, I am not a FORTRAN or ruby expert, but I understand they have both evolved enormously since their initial release and someone who has not kept up will at least be at a disadvantage. I do follow MS SQL Server closely and someone who learned on say MS SQL Server 6.5 would miss out on a lot of advantages of MS SQL Server 2008 if they had not made the effort to keep up with the evolutions in between.

As a side note, I knew a lady like that. She was a nice person, but she learned SQL Server 6.5 just to get a job and did not keep up with the changes. A lot of the code she wrote had to be run in compatibility mode and of course she missed out on newer features that would have made her code faster and easier to read simultaneously. She ended up getting laid off.

As for Cobol, correct me if I am wrong, but very little new development is done there, it is mostly maintenance coding and small upgrades for very old systems. Even if it paid the bills, I would not want to wind up as that guy maintaining ancient code that was simply too entrenched to be replaced yet personally.

Some of those Cobal people bill 500$ an hour which is worth putting up with a lot of pain IMO.

Unless these numbers are from a crystal ball and apply to circa 2015, it proves nothing about the relevance of Ruby skills a decade from now.

He's comparing old languages to a modern language.

Ouch, I completely misread the quote. My reply doesn't make any sense :/