[345tw4erfd]

IMHO a better way would be:

- A non-trivial take-home exercise and give them a week or two to complete. This should have higher weight on the decision instead of the "Google doc coding" as most likely that's the type of code they'll be shipping to production.

- Use the onsite interviews to improve upon the exercise and/or to get into the nitty gritty details, and also to make sure this is the type of person people would enjoy working with.

- Allow candidates to run the code and to look things up (even Einstein didn't remember how to do long division, he looked it up).

- Give people the benefit of the doubt and assume they are not liars or thieves. If you end up having certain doubts, ask yourself why you have those doubts and take appropriate steps to remove doubts.

And let's be real. Do you think someone with X years of experience having worked at multiple companies (small and big corps) was hired because they couldn't code?

[jwmerrill]

> even Einstein didn't remember how to do long division, he looked it up

Nitpick: the idea that Einstein struggled with school-level mathematics is a myth[1]. I would be very surprised if he ever had to look up the procedure for long division.

[1] http://content.time.com/time/specials/packages/article/0,288...

[ptd]

I personally hate take home tests that you aren’t compensated for, but I see your point. As to your last point, depends on your definition of “couldn’t code”. I think we tend to give the hiring and HR at companies like google too much credit. Things still slip through the cracks, especially at an organization of that size.

[345tw4erfd]

One thing I do with take-home exercises after I'm done interviewing with a company is adapting them and then open-sourcing them (obviously removing identifying information about the company). This has really helped me and I have more side projects to show later on. With that said, I've gone above and beyond with the implementation of these exercises so typically many of them are worth open-sourcing.

[oliTwist23]

Great Point. This is the right approach,as I have come to realize.

[joshuamorton]

People can cheat on take home exercises. At the scale of google, a nontrivial number of people will.

If you'll end up having to do a face to face analysis to make sure someone didn't cheat on a big take home assignment like that, why not just skip to the face to face analysis?