Its no that difficult to create a github project that can be completed in a weekend. I created the hacker new karma tracker and weedprices.biz in one weekend each.
But would you really make hiring decisions based on either of those projects? Software development is 'relatively' easy when your project is <1000 LOC. It becomes harder as the project grows. And it's the ability to grow the project what you should make the hiring decisions on.
Yeah but judging someone's ability to code by several good <1000LOC projects is probably better than the sorts of coding problems you have time to give someone during a coding interview.
That's actually a good point I have not though of before. I considered your Github account to be the place where you'd show off your l33t sk1llz, as opposed to show off code similar to the code you'd write during interviews.
Sure but then if this becomes a "must have simply to have the opportunity to interview" then it loses all the value it can have to reveal what the person can do. And I do believe some people have no time for this. Having worked at Apple for some time, I guarantee my sparse time, when there was some, was more outside of any coding and still I will never show outside a piece of code I have written in this company. This does not mean the coding skills I have should be considered as good or sufficient, they may or not match the one needed for a position. Having been also in the position to be on the interviewer side, especially in startup, seeing the candidate in person reminds often the deciding point.If some code on github exists, it simply can "speed up" a little this part of the process.
Personally I found the "Smart and get things done" book by Spolsky a good read on how to conduct interviews
