Most Effective Way to Evaluate Talent

I'm a contractor that is often looking for subcontractors, but I feel like there is way too much trial and error in the process. How do you screen talent, and what are effective ways for discerning between those that talk the talk and walk the walk?

Talent is hard to determine from just code samples and an interview - what innovative ways have worked for you?

From stackoverflow danpickett

• Get the candidate to write some code during the interview.

  From Galwegian

• I find that passion is a good measure. See if they really care about computing, of if it's just another job. A passionate developer may well have a blog or website, as well a participating in open source projects. If they are an open source developer, you can judge their code, design and often communication skills by looking at the project.

  See how interested they are in picking up new technology, not necessarily for immediate practical use, but just to stay on top of things.

  EDIT: Just to be clear, I mean these things to be in addition to normal questions and tests used to judge basic competency.

  From Jon Skeet
• 1. Do they spend time helping others on places like stackoverflow? If so, you can use that to gauge whether they can "walk the walk"
  2. Do they have a blog? A person's blog is often a very good measure of their technical ability?
  3. Do they contribute to any open source initiatives? If so, in what way? Take a look through the SVN (or whatever) repository for that project to see what their development skills are really like
  From David Arno

• if they interview well and you think you can work with them then often this is enough. People can spew out blog postings (and stackoverflow answers) all day long but still underperform, so look first for compatible personalities and working styles.

  From Steven A. Lowe

• Evaluate their ability to abstract, encapsulate, hide data, design for expansion (i.e., OOP skills if the job requires). Most evaluations focus on language familiarity and algorithms/data structures. These topics are rarely studied.

  When you have contractors, the main thing you need to know is how sandboxed you can keep their work, and those skills are vital.

  From Uri

• I have candidates whiteboard a solution for me to solve on an issue that may have been difficult for me or my team to solve. This gives me the ability to see if someone can think on their feet, and how they communicate and analyze. If they can come up with solution that matches or is better than what my current staff has formulated, I know that I have met someone who can supplement the team nicely.

  From David Robbins

• I'm curious - why is this getting down voted? Why do people feel that this question isn't relevant?

  For those of you that downvoted, please comment I'm interested. I think this is a huge problem in our industry.

  From danpickett

• These might help, there's loads more if you look -

  http://stackoverflow.com/questions/114950/how-do-you-verify-the-work-and-experience-level-of-consultants

  http://stackoverflow.com/questions/83185/should-you-administer-iq-tests-to-programmers-during-the-interview-process

  http://stackoverflow.com/questions/69238/job-interview-process

  http://stackoverflow.com/questions/58107/how-to-find-that-rock-star-junior-developer

  http://stackoverflow.com/questions/20555/are-written-tests-useful-skill-assessments-for-developers

  http://stackoverflow.com/questions/19801/what-interview-question-weeds-out-bad-applicants

  http://stackoverflow.com/questions/3568/how-do-i-interview-a-senior-developer-without-pissing-him-off

  http://stackoverflow.com/questions/196470/what-are-the-traits-that-you-look-for-in-a-software-developer

  http://stackoverflow.com/questions/110522/how-can-you-measure-your-skills-as-a-programmer

  http://stackoverflow.com/questions/106917/what-do-webdeveloper-employers-look-for-most-in-website-contracting

  http://stackoverflow.com/questions/25895/what-kinds-of-code-would-you-want-to-see-in-a-portfolio

  http://stackoverflow.com/questions/110974/what-employers-expect-from-new-developers

  http://stackoverflow.com/questions/57161/finding-the-right-consultant

  http://stackoverflow.com/questions/3247/identifying-passionate-programmers

  http://stackoverflow.com/questions/130759/what-skills-do-you-think-are-essential-to-becoming-a-jedi-programmer

  From Kev

• I agree with you that this is something that is very hard in our industry. It is increasingly getting more difficult for reasons beyond my knowledge. Therefore, I don't think you're going to find a simple answer. As you look around this site you'll see that different renditions of this question has been tackled before. All I have to offer is the following so please be gentle... :)

  In my personal experience, a programmer can be an absolute STAR in an interview but turn out to be a horrible disaster once hired on. The reverse is also possible, where the programmer might not be able to communicate effectively with text book interview questions, but in a real world scenario they are priceless!

  What this boils down to is that interviews are hard because they don't tell the whole picture. It seems that anymore, you have to know somebody, who know's somebody else... who recommends this person because of whatever reasons.

  I guess this translates to the only sure fire way to know whether somebody is good or not is:

  • You've worked with the person before
  • Or... you trust somebody that has worked with the person before
  From Scott Saad

• I didn't vote it down, but I would have.

  The issue is NOT that this isn't relevant, it is that it has been discussed to death already. Kev linked to 15 questions on this site alone. Joel Spolsky (one of the SO founders) has done a TON of writing at his site (http://www.joelonsoftware.com/) on this same topic as well.

  Voting down is the polite equivalent of "RTFM" or "Do you know what the search box is for?"

  From tomjedrz