Top Mud Sites Forum - View Single Post

thelenian · 07-22-2002, 06:22 PM

Guys, guys... the applicable concept in this case is KISS. You're accepting data from untrusted anonymous unidentifiable sources. There is simply no way to uniquely identify a given person. At best, you can reliably identify unique ip addy's, unless you can get ahold of the NIC MAC addy of each comp connecting, at which point you can uniquely identify each NIC (which means uniquely identifying a computer in most cases). Given this scenario, it is literally impossible to try to come up with a system that will reliably rank sites based on the number of unique people who actually play a MUD.

Tying votes to accounts rather than IP addy's simply adds another layer of redirection for vote cheaters. They just register multiple accounts. If you also bind accounts to unique IP addy's, you break people who use dhcp and/or dialup connections. Add detection and handling of dynamic IP address spaces and you effectively reduce the 10 people who play MUD X on the same dialup ISP to 1 person with respect to voting.

You could also try using cookies to uniquely id each comp, but all that means is that you need to delete the cookie before you vote.

Trusting MUD admins to accurately report their pbase size when that report directly affects their weighted ranking is the worst possible course of action. This is placing trust in the most untrusted part of the system. Also, weighted rankings with a max rating (i.e. 100%) are bad mojo, as you'll end up with around 50 MUDs tied at 100%, at which point some secondary criteria becomes most important, eliminating the purpose of the rating.

Heck, you could tie it all together, with unique IP detection, cookies, and account binding. What do you get? Tons of players who dialup and vote once each with Opera, Mozilla, Netscape, Emacs, Lynx, Galeon, and Nautilus (possibly I.E. if they're 'doze people), and whatever other browsers people have, each registered with a single account. With weighting votes with respect to reported Pbase size, an admin could simply report a size of 1, only vote him/herself, and not put a TMS voting link up on their website, nabbing the MUD a 100% weighted rating.

Basically, because of the premise of untrusted, anonymous, unidentifiable sources, designing a cheat-proof, or even effectively cheat-resistant system is impossible. Also, as you increase the complexity of such a system, the advantage shifts even more heavily in favor of the cheaters, as they are the ones who will go through greater lengths to understand (and therefore identify the possible abuses of) the system.

IMO, the system is fine as it is. Just take the rankings with a grain of salt (as is the case with any ranking system).

07-22-2002, 06:22 PM	#17
thelenian Member Join Date: Apr 2002 Posts: 122	Guys, guys... the applicable concept in this case is KISS. You're accepting data from untrusted anonymous unidentifiable sources. There is simply no way to uniquely identify a given person. At best, you can reliably identify unique ip addy's, unless you can get ahold of the NIC MAC addy of each comp connecting, at which point you can uniquely identify each NIC (which means uniquely identifying a computer in most cases). Given this scenario, it is literally impossible to try to come up with a system that will reliably rank sites based on the number of unique people who actually play a MUD. Tying votes to accounts rather than IP addy's simply adds another layer of redirection for vote cheaters. They just register multiple accounts. If you also bind accounts to unique IP addy's, you break people who use dhcp and/or dialup connections. Add detection and handling of dynamic IP address spaces and you effectively reduce the 10 people who play MUD X on the same dialup ISP to 1 person with respect to voting. You could also try using cookies to uniquely id each comp, but all that means is that you need to delete the cookie before you vote. Trusting MUD admins to accurately report their pbase size when that report directly affects their weighted ranking is the worst possible course of action. This is placing trust in the most untrusted part of the system. Also, weighted rankings with a max rating (i.e. 100%) are bad mojo, as you'll end up with around 50 MUDs tied at 100%, at which point some secondary criteria becomes most important, eliminating the purpose of the rating. Heck, you could tie it all together, with unique IP detection, cookies, and account binding. What do you get? Tons of players who dialup and vote once each with Opera, Mozilla, Netscape, Emacs, Lynx, Galeon, and Nautilus (possibly I.E. if they're 'doze people), and whatever other browsers people have, each registered with a single account. With weighting votes with respect to reported Pbase size, an admin could simply report a size of 1, only vote him/herself, and not put a TMS voting link up on their website, nabbing the MUD a 100% weighted rating. Basically, because of the premise of untrusted, anonymous, unidentifiable sources, designing a cheat-proof, or even effectively cheat-resistant system is impossible. Also, as you increase the complexity of such a system, the advantage shifts even more heavily in favor of the cheaters, as they are the ones who will go through greater lengths to understand (and therefore identify the possible abuses of) the system. IMO, the system is fine as it is. Just take the rankings with a grain of salt (as is the case with any ranking system).