Excellent advice. I work for a county planning agency, and back in the pre-PC days we conducted a couple of "assessment center" evaluations, which were all-day affairs where job applicants went through a series of in-office exercises that simulated the real work tasks they'd have to undertake if hired. They were laborious things for us, the evaluators, to plan and carry out, but we never selected an inept employee when using this process. In contrast, we've been cursed with numerous clunkers selected through the traditional written exam-oral interview process.

A practical evaluation needn't be an all-day affair; I've also sat on interview panels where essential map-reading and interpretation skills were tested on the spot with a short task given during the interview. There's no reason why basic computer skills couldn't be tested in the same way, as you've suggested.

Real behavioral interviewing. What a novel idea! Ask the person to do a simplified version of the responsibilities you are hiring for. This sure beats the approach that goes under the name of "behavioral interviewing" of asking people to self-report their accomplishments and activities. "Don't tell me. Show me." I've used this approach with quite good results. As you implicitly showed, the interviewer's preparation for the interview is key. You have to do a first class job of setting up the problems that you want the interviewee to solve so that you are testing the things you intended to test.

I would suggest to look for this type of test at a community college. Inquire with the instructors at the CS department. Probably with an instructor that teaches computer applications or intro to computer software or OS classes. Also, temp agencies probably have basic computer competency tests although I'm not sure if they would share them but it may be worth a shot.

I have 18+ years of PC skills and can't get an interview. This young pup doesnt know how to turn on a PC and he gets hired (albeit conditionally).

Sheesh!

Take a look at the International Computer Driving Licence (ICDL). It is intended to address exactly this need.

Several years ago I developed some tests similar to what you describe:

1) a Word test / typing speed test: I print a paragraph that describes our products and lists several tasks. I also warn the candidates in advance that they are being timed. I tell them the benchmark (depends on the job). The tasks are simple – change the font, number a list of four products, spell check, save the file to a specific location, etc.

2) an Excel test. This one has more steps, all “basic”: copy/paste, cut/paste, sort, insert, number, sum, save, etc.

These tests quickly expose those who lie or “fluff”. Some candidates tell me they are not proficient; if not critical to the job (otherwise, end of interview) I help them with the test, set the benchmark differently, etc.

The tests let me observe how the individual responds to stress and how he or she responds to a new task. Do they stare blindly, search frantically, make up excuses, sweat profusely, or ask for instruction/help/teaching? If the test is very easy for them, how do they react / behave? Are they indignant, understanding of the process, boastful, etc?

I learn more in the 5-10 minutes I spend with them then the other interviewers do in 30-45 minute conversations.

PS. For our engineers, our engineering manager has developed a 1-2 hour CAD and logic test that is part of their interview process (if they pass my tests).

Without knowing the business the recruiter is in, it is hard to answer this question:
Is it easier to teach the business, or the PC skills?

In most industries it is easier to teach basic PC skills to someone with a natural talent / understanding of the (non-IT) business.