I have a technical degree, but I tend to disagree. There is a good deal of pure signaling involved, but graduates of humanities programs on average can write and research better after they graduate then previously. Heck, most can read or write better than my CS-degree-having colleagues. I also think there's considerable value to society produced.
In any case, it's easy for companies to solve that problem, if they see an inefficiency. If IBM thinks people without degrees are an underutilized/undervalued labor pool, they'll start hiring them. And if the market stops demanding degrees, people will stop getting them. Until then, it looks like a demand-side problem to me.
I'm confused what you disagree on - do you disagree that signalling is wasteful, or that the skills gained (and later forgotten) in "any degree, we don't care which one" are economically valuable?
By the way, a great way to test whether education is human capital formation or signalling is to consider the consequences of failing and forgetting. If you flunked out of college, would you have the same economic outcome as if you passed and then forgot what you learned? If so, then college is human capital, otherwise it is wasteful signalling. (Hat tip: Bryan Caplan.)
If IBM thinks people without degrees are an underutilized/undervalued labor pool, they'll start hiring them.
Unfortunately, in big companies, this is tricky due to agency costs. The hiring agent has different incentives than the company as a whole - if he does something non-standard, he will receive disproportionate blame if it fails, while the company will gain most of the benefits if it succeeds. If he conforms to the standard procedure he can avoid such risks.
This is a fairly well known problem in the economics of large organizations.
FYI, I ignore degrees when I hire. But if I worked at IBM, I might not.
I think the mistake you're making is you assume it's the skills (gained and later forgotten) that the employers care about, but in reality the college degree is just a filter, not unlike an IQ test. Sure, it sucks that it takes four years and costs $100k, but hey it's not the employers that pay so what do they care. (That's why degrees are less important for engineers, since you can filter eng candidates by asking them a few coding questions.)
(Not that this makes the situation any less deplorable.)
It's very interesting to read your second paragraph and then follow it up by reading about Hungry Academy: http://hungryacademy.com/
Essentially, Living Social has done just this. They decided that the talent pool was thin, and the usual means of finding new hires weren't cutting it, so they mixed a bit of unconventional hiring with a dash of unconventional education and came up with the Hungry Academy. It will be interesting to see if this is a continuing trend, though I can't say I exactly relish the idea of corporate sponsored higher education as the new norm.
In any case, it's easy for companies to solve that problem, if they see an inefficiency. If IBM thinks people without degrees are an underutilized/undervalued labor pool, they'll start hiring them. And if the market stops demanding degrees, people will stop getting them. Until then, it looks like a demand-side problem to me.