That surely depends on whether the startups that fall through the cracks of the existing system caught by ApplyHN prove to be, on average, [close to] as successful as the ones picked up by the existing system and thus deserving of being caught. The cracks could be a feature rather than a bug.

Obviously you need more than two startups to make this sort of deduction, and I'm perfectly willing to believe that these and other startups missed by the traditional application process genuinely are more promising than some of those selected by it, but I'm not sure that what we have at the moment is really evidence either way.

Also very possible there's a bit of selection bias, since it wouldn't be surprising to discover that YC cross referenced the Apply HN apps with the YC/YCF apps and the act of doing that alone would be a source of bias.

This might be the case, but I think we should wait to make a final decision if it is a good way to catch startups that fall through the cracks. We really need a random selection to know if HN (and YC) are adding alpha.