I think the tricky brain teasers are fun, but when you use them, you run the risk of missing out on someone who might be a good developer just because they can't guess the trick. You also might get a bad programmer who just gets lucky and guesses the trick. These questions use up using a lot of time and don't necessarily tell you much about a candidate.

I disagree with a lot of the comments that say whiteboard problems don't tell you anything about candidates. I think seeing a candidate write code can be very helpful, as long as your evaluating problem solving skills rather than esoteric API knowledge. A good programing problem that requires several steps to complete tells me a lot about a candidate when I watch them solve it, even if they don't get the correct solution. I've started doing computer-based whiteboard problems which let candidates take advantage of an IDE with code completion and real-time error finding. This is more interesting, but you need to use a simple, well-known IDE (like Eclipse for Java) that someone can use easily without too much learning.

I agree with the OP that you never know how a developer will turn out until they do some real work.

There are 2 types of tech jobs:

1. Jobs where you're expected to do whatever it takes and you'll be rewarded with promotions and money for your extra effort
2. Jobs where you're not expected to go beyond the job description

I used to work for a major public university as a "Programmer/Analyst". I worked there for about 6 years. This was a Type 2 job. I did a lot of outside-the-box thinking, worked my ass off on projects that no one asked me to do, etc. My boss really liked me and encouraged me to do more. He gave me some small raises and job title bumps, but his hands were pretty much tied by university job descriptions and pay policies. Public universities don't have stock options. This would have been a sweet job if I just wanted to show up for work and enjoy the steady paycheck and the great health plan. But I wanted more challenge and more money, so I found another job.

A few years later, I found a job as a software engineer at a semi-startup SaaS company. This was a Type 1 job. I did a lot of outside-the-box thinking, worked my ass off on projects that no one asked me to do, etc. Executives recognized my efforts and abilities. Over the course of the next 7 or 8 years, I was promoted up to my current position as the head of software development. I really like my job. I don't work much more than I did at my university job, but the work I do is recognized and rewarded.

If you want a Type 1 job, find a tech company that's not too big with interesting products and a promising future. If you want money, make sure the salary is high and that you're getting good equity. Hopefully, your efforts will be rewarded. If they're not, go look for a different job.

If you want a Type 2 job, stay where you are. Or, go find a job as an SA/DBA/IT-guy at a university, law firm, medical practice, manufacturing company, or auto-parts reseller. Enjoy the steady paycheck and the good health plan. For a challenge, work on FOSS outside of work hours, and don't try to sell the results back to your current employer. Whatever you do, don't try to turn a Type 2 job into a Type 1 job. That's what you're trying to do now, and I guarantee that it will end badly.