It's a very interesting topic. I've worked at a company where we would get a stack of applications (about 1500) over night (right before the deadline of course) and the crappy tracker system we used would crash. Even worse I searched for a reasonable alternative, but simply couldn't find one that was SIMPLE. Sure there're plenty of bloated CRMs out there, and even some honest efforts like jobvite and interview street. But I couldn't find the right balance of simple tracking and review/screening features. So, as PG once suggested, I went and built my own.

It's still in dev mode but we've gotten pretty good responses from potential customers so far. Some of the issues we've tried to tackle has been the resume "problem" as discussed here. The thing is, resumes are much more than the content they hold. The design of the page, the choosing of words and descriptions, how to weight the important stuff etc. You can tell a lot about a person by looking at their 'paper' resume. It's almost like judging a person by their handwriting, though this metric is sort of skewed since everyone I know, myself included, has developed crummy handwriting ever since spending 10+ hours/day communicating via a keyboard.

An interesting point-of-view I've seen a lot of places is the obvious, yet under-prioritized factoring of values and soft skills. Personally, I think it's naïve to think you can attract "the best, the top 1%" of coders, managers, product devs and so on. What would be far more valuable in terms of hiring would be to find those people that would actually fit the company culture and team the best. Skills can be learned, but personal values and interests can't. So a silly manager might ask for someone under 30 with 15 years of experience in C#, Java, Ruby, and some html/css for good measure. Yet, this is the skewed metric to evaluate on. Instead, they should find the person who finds the problem they're meant to solve interesting and someone who would fit well into the company and the team. If the candidate 'only' knows C++ and Python, it's far easier for them to get up-to-date on the company program running on Ruby than trying to force the 'ideal hard skill candidate' into a culture he/she's not comfortable in. Prioritizing candidates in terms of soft skills and values are actually one of the pillars of the program I'm working on. It's a tough nut to crack..

Another issue we've talked about is how to evaluate someone with a minimum of screening bias? There's so much bias going into an evaluation of a person, and it's a well-known issue that we tend to look for people similar to ourselves - even though it might not be what the company is actually looking for. So, if the application tracker could enable "blind" mode and switch off certain information like pictures, names, tests scores etc. it could potentially negate some of the bias involved in screening. Even letting multiple people rate a candidate could negate some of it, yet the work flow used in most HR departments simply don't allow for this kind of flexibility.

Long rant, interesting topic.