Discussion over SRM difficulty
I love TopCoder and have competed in SRMs for over a decade in the lower end of Division 1. In this time there's been one issue that has never been addressed; the difficulty seems objectively too difficult.
Naturally we want to ensure that each competition has a range of difficulty so that both casuals have some engagement and success but plenty to strive for, and that the strongest are still challenged.
In general I would think a problem sheet with 3 problems might aim for the first to be solvable by 90% of entrants, the second by 50%, and the last by 10%. Additionally, points awarded for speed and for challenges, so the top people have plenty to compete over.
Looking at SRM 849, of the 42 registrants who opened at least one problem, 43% solved the first, 34% solved the second, and 0% solved the third.
So as things stand, half of entrants (already quite a veteran group) can't even make the "Easy" standard. Clearly the calibration is way off and needs more than a minor tweak. If you're at the lower end of Div 1, Blue-rated, only 2 of the 16 entrants could make it (and they probably aren't Blue any longer). Should a Blue-rated person be excited to join for SRM 850 with a tiny chance of hoping to solve any problems? There's such a thing as setting a high bar and hoping people rise to it, and there's such a thing as having such a high bottom rung that people leave to find something with a ramp they can feasibly ascend.
This SRM is just one example of the general trend; certainly not targeted at this problem set/setter in particular. He/she has set them in line with most previous SRMs, and I have enormous appreciation for the efforts and talents of the setters and TopCoder organisation. I write this because I believe TopCoder is an amazing and underappreciated platform which could grow in popularity again with a slightly more forgiving structure. Very keen to hear the thoughts of anyone who agrees or disagrees.