This is a detailed explanation of the format of my IRC usage stats. This covers the new version, which came into effect 04/96 or so. Report covers the period from 05-17-96 00:15 to 05-23-96 23:45, CDT 249 data point(s) taken! Pretty self explanatory. Basically, each network is polled twice an hour. So, for this report, the network was polled 249 times (sometimes it's not able to do it's poll, so this number can and will vary.) Average/peak number of clients on IRC during this period : 17781.6/22372 peak was reached at 05-20-96 22:15 CDT ... self explanatory. (1) (2) (3) (4)(5) (6)(7) (8) (9) #cli/ p c up% * /p* lnks/pl su% bd% server ------/---- --- ----/-- ----/-- --- --- ------ 916.3/ 951 98 5.3/ 7 7.2/11 22 20 irc.cerf.net 821.8/ 986 94 2.8/ 5 2.6/ 5 23 0 irc.ais.net Ok, here's where we get into the nitty gritty. (1) average number of clients that were on this server during this period. (2) like (1), but the peak number of clients. (3) the % of the time that this server was seen to be `up' and connected to the rest of the IRC network. This is somewhat unfair to remote IRC servers, but the error is typically small. (4) (5) the average and the peak number of IRCops seen on this server respectively. (6) (7) the average and the peak number of servers linked to this server respecively. Note that these numbers will be understated if the server is behind a hostmask - basically, any servers that are under the hostmask will not be counted. Hostmasks are listed later ... For a leaf that's never behind a hostmask, this number will be 1.0. If it is behind a hostmask, it could go down to zero. (8) % of users on this server that are on `single user machines'. For the most part, this means people on SLIP/PPP accounts. My program uses a list of dialup servers to determine if a machine is a single user machine or not. The file `clone.cfg' contains the list. `single:' lines indicate a regex (perl style) that refers to a single user machine. `single-override:' lines override this. single: .dial-access.att.net$ single: .hsk.fh-sbg.ac.at$ single-override: dismayl.demon.co.uk$ Other lines, like ignore: are not used for these statistics, but for other programs of mine (clonebot detection.) Feel free to give me more information to put into this list. (9) % of users on this server with broken DNS - basically, rather than appearing to be on the host `piglet.cc.utexas.edu', their machine is given as an IP - `128.83.42.61'. Note that my program does not take into account it's own DNS lookups for this number - this number is `raw'. This number may or may not be useful to anybody :) --- Part 2 - hostmasks seen --- Server hostmasks seen : *.au(41%) *.be(97%) *.netcom.com(33%) *.portal.com(99%) *.de(7%) *.rutgers.edu(86%) *.uci.edu(97%) *.fi(58%) *.inet.fi(6%) *.lut.fi(29%) *.oulu.fi(4%) *.hu(35%) *.ac.il(87%) *.it(98%) *.jp(96%) *.epix.net(97%) *.stealth.net(4%) *.texas.net(100%) *.no(97%) *.se(98%) *.si(97%) *.tw(25%) This section lists the hostmasks seen during this report period, and the % of the time each hostmask was seen. This is helpful in determining how accurate the links/peak links numbers in part number one are. --- Part 3 - stats by second-level domain --- (1) (2) (3) (4) (5) (6) (7) (8) (9)(10) # cli/pcli n-dup/nd-p d % su% * /p* s ps domain ------/---- ------/---- --- --- ----/-- ---- -- ------ 1188.7/2215 1096.8/2017 8 92 1.3/ 5 2.5/ 5 *.netcom.com 946.0/1245 796.2/1055 16 0 2.0/ 3 0.0/ 0 For this part, my program attempts to fix problems with broken DNS. If a client shows up as coming from an IP address, it does a reverse DNS lookup. If this reverse DNS lookup fails, or returns a bogus value, the client is lumped in with the domain. (1) (2) average, peak number of clients seen from this domain. (3) (4) average, peak number of `unique users' seen from this domain. This is a bit complicated. Will cover later. (5) basically, this is `( (1) - (2) / (1) ) * 100 %' Some people like to call it the `bot %', but that is not really accurate. It lists the % of clients that are `duplicates', be they bots, or a person with two windows open. (6) % of users in this domain that come from `single user machines' (see above.) (7) (8) average, peak number of IRCops seen from this domain. (9) (10) average, peak number of IRC servers seen in this domain. --- Part 4 - stats by top-level domain --- (1) (2) (3) (4) (5) (6) (7) (8) (9)(10) # cli/pcli n-dup/nd-p d % su% * /p* s ps domain ------/---- ------/---- --- --- ----/-- ---- -- ------ 4717.7/7255 4265.7/6655 10 42 42.8/55 15.3/19 *.com 3725.4/5758 3296.2/5208 12 37 37.5/50 21.1/23 *.net This part is exactly like the previous part, except it refers to the top level domains instead of the secondary domains. --- Part 5 - Server `map' --- uni-linz.ac.at : irc.wu-wien.ac.at(76%) *.wu-wien.ac.at : sil.polytechnique.fr(53%) irc.nada.kth.se(2%) irc.wu-wien.ac.at : uni-linz.ac.at(76%) irc.felk.cvut.cz(4%) cismhp.univ-lyon1.fr(24%) *.hu(2%) *.au : irc.blackened.com(3%) irc.stanford.edu(37%) ircd.stealth.net(2%) This part just lists which servers linked which other server, and the % of the time that this link was active. Note that it makes no distinction whatsoever about who the `uplink' is and who the `downlink' is.