In orderthe early days, the ''Iron Man'' system was set up under which coders were grouped into rooms of ten according to encourageskill newcomerslevel (determined by [http://www.topcoder.com/index?t=support&c=ratings rating]), in theorder earlyto daysencourage newcomers (who would have little hope of beating out the best coders for prizemoney) to stick around and compete. After the contest, the top (3?) (highest scoring) coders in each room were paid according to the skill level of the room (winners in the room of contestants with the top 10 ratings were paid more than those in the room with coders 11-20, and so on). This created some paradoxical situations such as the possibility of coming fourth in the whole contest and getting no money, while the 111th placed contestant got paid, and gave rise to ''ratings diving''. Ratings diving, or taking a ''ratings dive'' was accomplished simply by doing very poorly on purpose in a particular contest (by opening and not submitting any problems or submitting incorrect challenges). As a result, a contestant would, in their next contest, be placed in a room with 9 coders among whom they had a good chance of winning some money. The idea was that the dramatically increased chance of winning money made up for the fact that the money to be won was a lesser amount given the lower average rating of the room.
Since, coders have been divided into two divisions (appropriately named Division I and Division II), and are grouped in rooms of up to 20 in such a way that within each division, the average coder ratings in each of the rooms are roughly equal.
