OK, I understand. Well, one idea has just occurred to me now. We can make something like the statistics. We can observe and monitor what is the average ratio between all and well-seeded swarms. I think it is fairly recurrent in a net. We can make several tests to find the best value. Then we will be able to create a better formula. For example: (after all our tests) the ratio is 10% (10 well-seeded for 100 peers) So the formula for the Maximum number of connected peers per torrent would be: MNOCPPT = (10 * all peers) / 100 Of course we also have to include all seeds as the Maximum number of connected peers per torrent is related not only to peers. It refers to seeds + peers. We can also modify this value and make it dependent on the max. upload speed. Do you think it is possible to find such a formula ? ... It may be possible, but math is really not one of my strong skills...so I'm clearly not the best person for that job. :| In addition, there are other fairly important variables that will skew the formula's results. Swarms, for instance, are very dynamic in nature. It's incredibly rare that you'll find a swarm that more or less stays roughly the same size and maintains the same seed to peer ratio. You have to take into the equation the relative speed of the whole swarm (this on its own is difficult to establish - if not impossible), the seed to peer ratio as mentioned previously (which constantly changes in most swarms) etc. etc. etc. Plus, you have to deal with the fact that if suddenly the swarm halves in size (maybe a tracker has gone down and xx.x% of peers do not have DHT/PeX enabled), BitTorrent users are going to have to change their formula to reflect this unexpected change, if they want to keep the most optimized settings for that particular torrent. IMHO, the enormous task of finding this formula, either through probability or accurately computing values based on seed to peer ratio and things like that, is really way beyond my scope. Don't get me wrong, I like your idea, but in practice I'm not so certain it'll yield the same satisfactory results as in theory. That's my 0.03¢ (damn inflation).