I agree with most of what you said, but one way you can lose if you insist on contract-to-hire is that you are filtering out a portion of the highest quality candidates who are only interested in immediate full time hiring.
No one desires to be in a contract limbo where they aren't sure if they're going to be hired for real or or if they're doing the job search all over again in a few weeks.
Further, as soon as you use contract work to determine hiring criteria -- you've added your management into the mix. Are expectations made clear? Is the person's time managed efficiently? If management fails at any point then it's detrimental to the prospective employee's future, likely not by their own fault.
The point is, there's no reason for it to be a genuine contract position, and every reason for it to be concise enough for them to fit in during their off hours. After all, you are going to have to look at it too :)
And they do have the time for it: after all, they are making the time for interviewing.
It also gives the candidate a way to see what kind of problems he'd be working with, and the quality of the code-base, working conditions, etc.
I think it's a win-win.
Trouble is most employed people will have restrictions on outside work and if its related work the current employer will own that IP.
One size definitely does not fit all here.
Quitting a full time job for a new position is a big commitment on my part, and I'd like a big commitment on the other side of the table. Not to mention contracting gigs don't usually have insurance, vacation, or other benefits.
If it means a more rigorous interview that's okay with me.
An unemployed candidate can do a contract like that while evaluating other jobs and possibly waiting for other offers to come in.
An employed candidate can do the work in off-hours.
if you're not able to easily spot such a candidate without attempting to make him/her to jump through the gimmicks (brainteasers, contract-to-hire, etc..) - well, it indicates the issue with your company and hiring practices in particular are deeper than just what gimmicks to apply.
I'm not sure if this is a joke, but I'll bite.
The entire problem with hiring is that no-one, I repeat, NO-ONE is capable of determining who are the best candidates. If they could then everyone would the same formula and this would be a solved issue.
GOOG and MSFT have tonnes of hiring data and they admit they can't do this very well. It seems very presumptuous of you to assume you can and anyone who can't is somehow an outlier.