00:00:04 I mean, they take floppy disks. But apart from that, they are not obsolete! 00:00:12 sounds like obsolescence to me! 00:00:54 -!- osaunders has quit. 00:01:20 I guess one of those player pianos was actually an electronic piano. 00:02:04 ive been tempted on numerous occasions to build an electromechanical computer 00:02:07 Which is pretty much the same thing as a keyboard, except it's actually fashioned to look like a piano and doesn't have as many features. 00:02:25 uorygl, my piano takes an usb cable 00:02:28 Ah yes, electromechanical computers. Have you ever designed one? 00:02:34 anmaster: "an usb cable"? 00:02:35 :| 00:02:39 augur, yes? 00:02:43 midi usb? 00:02:45 augur, it's called MIDI over usb 00:02:45 yes 00:02:48 AnMaster: its "a usb cable" 00:02:51 augur, why 00:02:54 AnMaster: is it a player piano, an electronic piano, or a keyboard? :-D 00:02:58 because it doesnt start with a vowel. 00:03:05 augur, you pronounce it "U S B" 00:03:07 not "usb" 00:03:10 you-ess-bee 00:03:18 augur, yes indeed 00:03:20 the name for the letter "u" does not start with a vowel. 00:03:27 augur, what about pronouncing it "usb" though 00:03:33 then "an" would be correct 00:03:38 yes. 00:04:01 augur, well then I pronounce it that way ;P 00:04:05 uorygl: partially. my idea was to try and imagine what a computer wouldve looked like if you took a morse-code like device and extended that model 00:04:21 My ideas have involved creating a computer using only relays. 00:04:22 -!- augur has quit (Read error: 104 (Connection reset by peer)). 00:04:27 * uorygl blinks. 00:04:33 -!- augur has joined. 00:04:36 My ideas have involved creating a computer using only relays. 00:04:44 yeah 00:04:48 uorygl, also, no one uses a player piano these days 00:04:50 it always seemed odd to me that we didnt have electromechanical computers 150 years ago given that we had morse code back then 00:04:51 long live midi 00:04:59 and hardware midi to be specific 00:05:01 AnMaster: I have heard of MIDI player pianos. 00:05:03 AnMaster: noone uses midi these days. long live mp3. 00:05:32 MP3 is inferior to MIDI when it comes to things MP3 is incapable of doing and MIDI is capable of doing. 00:05:38 augur, no one uses mp3, Long live flac 00:05:44 :P 00:05:46 and what uorygl said 00:05:53 the converse is true too ofcourse. 00:06:04 i think we should build an electromechanical computer. 00:06:05 I agree that we ought to have had electromechanical computers since 1835. 00:06:25 oh? 00:06:26 you know that the morse code machines were originally like type writers 00:06:27 ? 00:06:32 midi is the shit you use to make mp3 00:06:34 I think I had some idea. 00:06:39 madbrain, no 00:06:40 but some douchebag convinced morse to use a push-level machine instead 00:06:42 not quite 00:07:00 I think we should build an electromechanical router. Then we could make an Internet using only 19th century technology. 00:07:01 were it not for that, we'd probably have had the internet in 1850. 00:07:04 midi is a transport protocol 00:07:14 like IP, right? 00:07:21 think of it, an electromechanical type writer that could connect to a tape store remotely 00:07:29 unless you do music with straight musicians or something 00:07:31 lament, and a file format 00:07:42 i only do music with gay musicians 00:07:51 ? 00:07:55 the tape would store bits directly rather than as letters so it'd just feed right into the line 00:08:16 So, augur, let's figure out an error correction scheme that can be implemented using relays. 00:08:29 ECC? 00:08:32 well you wouldnt have error correction initially, right 00:08:35 you'd just have hard connections 00:08:49 augur, reed Solomon? 00:08:52 and it wouldnt matter much either, because the signals are pretty strong, comparatively speaking 00:08:53 Reed* 00:08:59 Isn't error correction pretty necessary? 00:09:00 AnMaster: whats that 00:09:03 augur, iirc 00:09:14 augur, http://en.wikipedia.org/wiki/Reed–Solomon_error_correction 00:09:16 uorygl: only when youre signals are weak 00:09:16 uorygl: depends on the noise levels and such 00:09:19 augur, used on CDs and such 00:09:31 Well, the signals are going to get corrupted every time you do something to them. 00:09:47 sure but an electromechanical computer is going to emply pretty strong signals anyway 00:09:48 augur, works by oversampling a polynomial, thus being able to reconstruct the missing data points if some are gone 00:09:55 so the signal degredation isnt going to be significant 00:10:02 If a signal goes 1,000 miles and passes through 10 routers, I think there's going to be quite a bit of signal degradation. 00:10:16 maybe maybe not, uorygle 00:10:22 ah, but that's transmission, not computation 00:10:26 Even worse if a signal is circulated for an indefinite length of time. 00:10:35 Besides, error correction is easy to implement, is it not? 00:10:47 -!- osaunders has joined. 00:11:14 i mean, they had routers back then for these signals 00:11:23 they already had transatlantic telegraphs 00:11:29 True. 00:11:31 by like 1850 or 1880 or whatever 00:11:37 so i dont think thats an issue 00:12:45 brb pizza :D 00:14:20 Okay, so we don't need error correction. 00:15:05 So, what sort of signals do we want to support? Packet switching? Circuit switching with in-band signaling? 00:15:19 How would packets be delimited? Time? Number of bits? 00:15:54 in band is bad 00:16:08 * AnMaster whistles 00:16:13 :-) 00:16:39 In-band is good! It means you only need one band. 00:16:50 The phone guys came up with filters that prevented the whistling stuff, no? 00:17:18 uorygl, so some freqs were forbidden? Would break said data 00:17:25 you need to *escape* it instead then 00:17:39 Now, let me read a bit about how relays work. 00:18:01 uorygl, well if you were going to build a computer with that you could do it high level anyway 00:24:20 It looks like in an ordinary relay, the switch is thrown by passing current in either direction. 00:24:26 o hai 00:25:17 So you could say that a packet ends whenever the voltage drops below the threshold. 00:25:25 * AnMaster locates a boot cd with gparted 00:26:16 Or you could use a latching relay, and say that the packet ends whenever the voltage becomes negative; then the data can include both positive and zero voltages. 00:26:20 uorygle, lets not try to rebuild TCP/IP on aethernet just yet 00:26:27 TEEHEE AETHERNET 8D 00:26:32 -!- anmaster_l has joined. 00:26:49 Well, the thing about routers is that they route data. I'm just pondering how that data could be delimited. 00:27:19 special bit patterns, obviously 00:27:37 That won't do if you're transmitting analog data. 00:27:53 special tone patterns, obviously 00:28:10 That requires something other than relays. :-P 00:28:26 analog data would in general! 00:28:28 uorygl, what about a fixed packet size? 00:28:48 x milliseconds before switching 00:29:04 I don't know if that's a good idea. 00:29:13 i think we should get our system up and running locally first before trying to get a transatlantic system up 00:29:22 I think the packet overhead could potentially be several seconds. 00:29:39 uorygl, also what about bouncing with relays 00:29:48 wouldn't it be a rather severe issue 00:29:57 We can compensate for bouncing by using error correction. >.> 00:30:07 uorygl, or Hg relays 00:30:09 read about that 00:30:13 no bouncing in them 00:30:17 of course they are toxic 00:30:24 which is a rather large downside 00:30:30 guys we should really build one of these 00:30:36 we could make it all steampunkish 00:30:45 and show it off on one of the steampunk blogs 00:30:47 augur, it will eat more power than my old p4, and it will also be slower 00:30:50 Let's each build our own and then figure out how to connect them. :-P 00:30:55 anmaster: EXACTLY! :D 00:30:58 itll be AWESOME 00:31:06 congrats, you get promoted to Intel Chief Engineer some years ago 00:31:16 HOORAY 00:31:30 augur, then you were fired when they decided to produce core 2 00:31:49 : 00:31:51 :( 00:31:52 :? 00:31:54 ah 00:32:05 btw, going to go offline with this connection for a while 00:32:07 parted time 00:32:18 parted time? 00:32:26 augur, yeah parted 00:32:30 what 00:32:34 augur, try it's man page 00:32:40 man parted 00:32:44 you type that in your shell 00:32:50 i'd part a man 00:32:53 IF YOU KNOW WHAT I MEAN 00:32:54 ;D 00:32:58 no I don't 00:33:02 oh 00:33:03 ok 00:33:07 Hmm. Packet switching and circuit switching aren't really that different when packets can go on for long periods of time. 00:33:19 GNU Parted - a partition manipulation program 00:33:20 augur, 00:33:22 was that so hard 00:33:27 read your damn man pages 00:33:30 AnMaster: i didnt care :D 00:33:35 uorygl, have you seen those little calculators 00:33:38 * coppro goes to play more engineer of the people 00:33:40 theyre like little black drums with a crank? 00:33:50 Now I'm thinking that everything should be digital and packet-switched, and then we can just use special bit patterns. 00:33:54 I don't think I have seen those. 00:34:40 http://www.vintagecalculators.com/html/curta_i.html 00:35:11 What do you call that code where every 0 is encoded as 01 and every 1 is encoded as 10? 00:35:20 dunno 00:35:29 weirdonary? 00:36:38 There's a name for it. 00:40:30 Aha. The Manchester code. 00:42:48 What's the poin.. oh, error detection 00:42:49 can we not use crazy coding schemes for this? :| 00:43:02 I wonder what a binary version of the L&S sequence would look like 00:43:09 L&S? 00:43:14 look and say 00:43:15 I guess crazy coding schemes aren't strictly necessary. 00:43:37 1, 11, 101, 111011, 11110101, 100110111011 00:43:37 We could just transmit at a constant rate. 00:44:07 coppro: that's kind of an irreversible sequence. 00:44:20 it is 00:44:31 I wonder at which base it becomes reversible 00:44:38 Try this: 1, 011, 010101, 010011010011010011, . . . 00:44:41 actually, no I don't 00:44:45 that's easy 00:46:05 I guess the thing about not using any code is that timing errors can happen. 00:46:20 Suppose they transmit 010101010101010000000000000000000000000000000000000000000000001010101010101010. 00:46:33 You're going to have to time all those zeros in the middle in order to know how many there are. 00:46:48 0 -> 01, 1 -> 10... NRZ? 00:46:51 Then again, we probably could use an error correction scheme wherein that can't happen. 00:47:00 Ilari: congratulations, you've invented the Manchester code. 00:47:37 now if you iterate it, you can invent the thue-morse sequence too :) 00:48:13 Well, the point of coding that mangles symbols is usually to avoid long runs of same symbol, as that tends to mess up timing... 00:48:21 Right. 00:49:00 Well, Manchester coding is one way to do things. Using it, I think we can pretty much correct for every sort of error we need to correct for. 00:49:11 i recall someone (ais523?) saying something about using 01 and 10 to ensure there were equal number of 0s as 1s to have no net charge 00:51:03 uorygl: um you cannot correct for an actual switching of neighbor bits :D 00:51:23 Bits don't just spontaneously get switched. 00:51:40 no, but you could have two neighboring errors 00:52:01 oerjan, hi 00:52:06 AnMaster: hi 00:52:15 oerjan, in a bit, once this CD is burned, you could say I parted to use parted 00:52:20 One of the routers could also spontaneously explode, causing a very large number of errors. Though that's less likely. 00:52:49 The thing is, Manchester at least has some error correction capability for every likely type of error. 00:53:08 oerjan, hey it's supposed to be a bad pun 00:53:19 AnMaster: yeah, yeah 00:53:29 oerjan, you know what parted is? 00:53:32 There are sequences over four symbols that never have any subsequence followed by any permutation (including identity) of that subsequence. Keränen's sequence is one of those (has recursive structure with generator of length 85). 00:53:39 yay done 00:53:54 i have heard the name, and can guess it's a partition editor 00:53:58 Ilari: I assume that means "immediately followed". 00:54:07 uorygl: Oops, right. 00:54:12 oerjan, indeed 00:55:05 -!- AnMaster has quit ("ZNC - http://znc.sourceforge.net"). 00:55:23 Ilari: um subsequences must have at least length 2, i take 00:55:31 Any sequence with recursive structure has to start with "identity" symbol. 00:56:13 oh wait fours symbols duh 00:56:18 *four 00:58:37 Let G be some group. Finite sequence a1 = e, a2, ... an can be expanded into infinite one by a1, a2, .. an, a2 + a1, a2 + a2, ... a2 + an, a3 + a1, ... an + an, a2 + a1 + a1, ... 00:59:04 Ilari: Keranen's sequence has "abc" at the start, and "bac" later on... or do I misunderstand? 00:59:15 coppro: Immediately followed 00:59:22 oh 00:59:31 (that was mistakenly left out). 00:59:43 ok, makes more sense 00:59:43 where did ais go 00:59:47 hm 00:59:57 There is no sequence over finite such that no subsequence eventually repeats. 01:00:19 *finite alphabet 01:00:28 right 01:00:36 I realized that about 2 seconds after I said that 01:00:43 Counterexample: "a" 01:00:44 :-P 01:00:57 Also *infinite sequence 01:01:08 uorygl: hah wrong, epsilon is both at start and end 01:01:20 Aww, you're right. 01:03:33 In Z2, using generator 011 would yield 011100100100011011100011011100011011011100100011100100100011011011100100011100100... or something like that. 01:04:14 Here, let me generate an infinite sequence. 01:04:36 ...hmm. 01:05:03 Yeah, I can do that. 01:06:02 Using elliptic-curve-type group with large amount of points could probably make some whacky sequences. 01:06:03 Using Thue, an excellent infinite sequence generator. 01:06:41 And of course there is: Monster group! 01:06:57 Aww, fudge. This Thue interpreter doesn't seem to be working. 01:07:23 Except that monster group isn't abelian, and thus one would need to define order to do the additions in. 01:08:17 Okay, here is an awesome sequence: 11011011101101101110110111011011011101101110110111011011011101101110110110111011011101101101# 01:09:08 Put all generators of monster group as generator of sequence and eventually the infinite sequence resulting would contain all elements of monster group. 01:09:22 That sequence consists of strings of 1s separated by 0s. Those strings have the following lengths: 2, 2, 3, 2, 2, 3, 2, 3, 2, 2, 3, 2, 3, 2, 3, 2, 2, 3, 2, 3, 2, 2, 3, 2, 3, 2, 2, 1 01:10:08 Ignoring the 1 on the end, that sequence consists of strings of 2s separated by 3s. Those strings have the following lengths: 2, 2, 1, 2, 1, 1, 2, 1, 2, 1, 2 01:10:23 And so on. 01:11:42 Unfortunately, my Thue program simply produces what is effectively an arbitrary finite piece of a random infinite sequence. It cannot generate a continuation of that sequence. 01:11:44 Of course, monster group has 808017424794512875886459904961710757005754368000000000 elements and duplicates probably exist before last element is found. Not to mention that computing group addition for moster group is quite slow. 01:12:24 -!- soupdragon has joined. 01:12:52 At least it can generate sequences of arbitrary length. 01:13:57 Monster group has generator of size 2. And interestingly monster group is isomorphic with subset of 196882x196882 matrices over Z2 with matrix multiplication as addition operator. 01:15:08 isn't every finite group isomorphic to such a subset, really 01:16:05 i suppose the interesting part is that 196882 is rather small 01:16:41 Of course, 38 762 521 924 elements total... 01:17:38 although the full matrix group is probably enormous compared to the monster 01:18:31 Well, the full group of that size would be all matrices of that size over Z2 with determinant 1. 01:18:50 so, unless there is only a tiny minority of determinant 1 elements 01:19:58 hm you can do it by choosing independent vectors 01:20:26 Well, there are already 2^19381162521 upper triangular (and same amount of lower triangular) such determinant 1 matrices. 01:20:28 first vector anything non-zero, so 2^196882-1 elements. that's already anormous 01:20:47 right, so enormous 01:20:53 *e 01:24:02 Hmm... Is Z3 isomorphic to some multiplicative matrix subset over Z2? Its equivalent to question if there is matrix over Z2 (and if there is, what's the smallest one) that has nontrivial cube root of identity matrix. 01:24:04 -!- coppro has quit (Remote closed the connection). 01:24:31 -!- coppro has joined. 01:24:33 every permutation group Sn is trivially embeddable into n x n matrix 01:24:45 *matrices 01:24:47 trivially ? 01:25:42 a permutation s is mapped to the matrix with M_ij = 1 is s_j = i, 0 otherwise 01:26:26 or is that s_i = j 01:26:34 -!- lament has quit. 01:26:54 -!- osaunders has quit. 01:27:54 well they're even called permutation matrices iirc 01:28:35 http://en.wikipedia.org/wiki/Permutation_matrix 01:28:38 Actually, any permutation cycle of length n can be used to find subset of matrices over Z2 that are isomorphic to Zn. 01:30:36 ah my confusion of whether to do s_i = j or s_j = i seems to have confused other before, because the standard definition is _wrong_ :D 01:30:43 *others 01:32:27 Actually, it seems that for any Zn, there is isomorphic nxn matrix subset over Z2 with multiplication. 01:32:44 yeah 01:32:58 every group is a subset of a permutation 01:33:07 and if you can matrixify permutations then you can matrixify any group 01:33:41 -!- MizardX has quit ("zzz"). 01:34:56 i think when n is a prime and using permutation matrices, that may be minimal 01:35:23 since p does not divide (p-1)! 01:36:05 may ask if it is still minimal if not using permutations... 01:36:47 (order of S_n is n!, and subgroup orders always divide total group order) 01:41:47 And if isn't minimal, fun question is what is smallest n where it isn't and what would be generator matrices for that? 01:41:52 if it is not a prime power then it is not minimal, because n = n1*n2 with n1, n2 coprime means Zn = Zn1 x Zn2 which can be embedded in S_(n1+n2) by using blocks 01:42:18 That would give n=6... 01:43:09 -!- adam_d has quit (Read error: 60 (Operation timed out)). 01:43:37 i'm not sure S5 is minimal for Z6, though 01:44:01 [[0,1,0,0,0][1,0,0,0,0][0,0,0,0,1][0,0,1,0,0][0,0,0,1,0]]? 01:44:16 S3 is too small, only 6 elements that _don't_ commute 01:44:18 -!- lament has joined. 01:44:32 i meant i was not sure S4 doesn't work 01:45:24 come to think of it there's probably a sequence in that sequence encyclopedia for this :D 01:45:37 If you have program that can calculate that, try calculating 2nd, 3rd and 6th powers of matrix I gave (over Z2)... 01:46:57 um i can see perfectly well it's using the block method i suggested, no need to use a program 01:51:11 Current methods would give 2, 3, 4, 5, 5, 7, 8, 9, 7, 11, 7, 13, 9, 8, 16, ... 01:52:09 hm wait of course the generator must be a _single_ permutation... 01:52:10 Oops, that should be 11, 8, 13... 01:52:41 hm that means this actually is optimal, i think 01:53:21 dividing that permutation into cycles is the same as dividing the matrix into blocks, and the resulting order is lcm of cycle length 01:53:35 so *lengths 01:53:59 Oops, 11, 7, 13... 01:54:06 so indeed dividing n into prime powers and adding them is the best you can do 01:55:05 "a(n) is the minimal number m such that the symmetric group S_m has an element of order n - Ahmed Fares (ahmedfares(AT)my-deja.com), Jun 26 2001". 01:58:03 And indeed element of order n means subset is isomorphic to Zn. 01:58:10 naturally 02:00:07 Hmm... I guess I should write my own raytracer to properly trace this fractal pattern. Pov-Ray can't do it properly because of 255 reflections limit. 02:00:34 YafRay? 02:01:55 Nah. Fairly simple to write raytracer that can just deal with it. 02:02:16 can you write a 'ray tracer' based on quantum physics? 02:05:04 so that it renders refraction and stuff 02:05:50 http://en.wikipedia.org/wiki/Field_with_one_element O_O 02:11:30 When I walked around at school, I couldn't help thinking that the tiles on some of the walls looked ray-traced 02:12:55 Why can't the additive identity and the multiplicitive identity be the same? 02:13:07 one of the axioms is 0 <> 1 02:13:17 Oh 02:14:21 -!- lament has quit. 02:50:33 It it particularly useful to say that there is no field with one element? 02:55:52 "For technical reasons, the additive identity and the multiplicative identity are required to be distinct." 02:56:10 * oerjan looks for actual reasons... 03:00:40 Once ReactOS is released and fully stable, would there be any real advantage to using it? I mean, it will be prone to Windows viruses, presumably 03:01:28 Although it would be nice that in 10 years after release, if ReactOS is still alive, it would still be considered modern, without having to pay money, unlike with Windows 03:02:31 * oerjan didn't really find any 03:02:54 Well, ReactOS will be free. We could extend it in ways that we can't extend Windows. 03:03:53 oerjan, because then it's not what the mathematical community decided is a "field"? 03:04:49 that is _not_ an improvement to "for technical reasons" 03:07:08 it's probably something to do with homomorphism 03:08:54 well it would be a terminal object... while other fields only have homomorphisms at all if their characteristics match. or wait, are there homomorphisms from Q to F_n? 03:09:28 *if at minimum their characteristics match 03:09:36 maybe it's for divide by zero 03:10:00 um wait no, obviously not 03:10:11 the characteristics must match 03:10:30 I don't know the characteristic ;/ 03:10:53 smallest integer n with n*1 equal to 0, or 0 otherwise 03:11:03 oh right 03:11:05 *positive integer 03:11:19 how is n*1 ever not n :S 03:11:57 well i just wanted to point out that n is not a field element 03:12:05 should it be n+1? 03:12:19 n*1 = sum of n 1's 03:12:27 oh!!! 03:12:32 so n is a natural number 03:12:34 it's "intuitively obvious" 03:12:54 it's not an element of the field 03:13:05 and in characteristic 0, there is no harm in identifying them, since the rationals are always embedded 03:15:58 n is always a prime or 0, btw 03:17:58 and for each characteristic n, there is a unique prime field, those are Z_n and the rationals 03:18:33 and the prime field would be an initial object in that subcategory, since it embeds uniquely in every other 03:19:10 the field of size 1 ruins all that 03:19:45 it would be of characteristic 1, i guess, but still have a homomorphism to it from everything else 03:21:17 myth busted! 04:13:49 Hey, now. I'm not following all that well, but it seems like you have certain objects associated with p^n where p is prime, and letting n be 0 ruins stuff because p^0 is a factor of everything. 04:19:36 on the other hand i think that may actually fit _because_ you always have a homomorphism to the 1 element field 04:20:18 and you may in general have homomorphisms from F_p^n to F_p_m when n divides m, or something like that (let me check) 04:20:24 *F_p^m 04:21:31 well you would _need_ n dividing m, at least, since then F_p^m is a vector space over F_p^n 04:22:00 if of dimension d, then p^m = p^(nd) => m = nd 04:27:29 * oerjan doesn't find an easy reference 04:37:11 -!- Asztal has quit (Read error: 110 (Connection timed out)). 04:55:02 -!- zzo38 has joined. 04:55:18 What should I name this class: http://www.dandwiki.com/wiki/User:Zzo38/untitled_class_1_(3.5e_Class) 04:57:00 ceci_n_est_pas_un_classe 04:57:35 heh 04:58:26 sorry, *_une_ 04:59:22 I like this joke but don't want to call it that. 04:59:43 i take it the improbability drive is the important aspect 05:00:19 It is one of the important aspects, but probably not the most important one 05:00:36 "dand"? 05:00:57 That's the domain name, I didn't write it 05:00:58 D&D 05:01:10 "it" is the domain name. 05:01:38 I guess it's better than "dandd". 05:01:38 well rephrase that: the name "improbability drive" was the only part i got any gist from 05:01:47 But is it better than "dnd"? 05:02:16 OK. But can you understand how all the class features works, however? 05:07:05 there is something called complex abilities. however i do not know d&d sufficiently well to understand how this is any different from what any spellcasting/psionics character can usually do 05:07:53 it's okay, oerjan 05:07:59 D&D 3.5 is not meant to be understood 05:08:03 heh 05:08:32 well let me leave it to the experts then 05:11:20 The "Complex Abilities" collectively refers to the spells, powers, and potentially others. 05:12:30 maneuvers? 05:13:45 "Complex Abilities" does refer to maneuvers, but whether or not I add that to this class is different 05:17:37 <3 LoZ music 05:23:16 One of the types of complex abilities that I might add, though, is warlock invocations, if I can figure out the cost that should be applied to them 05:37:40 just add them all 05:41:33 But what C.A.P cost? I want it to be not too low, and not zero like a actual warlock's invocations 05:41:49 And anyways, what should I name this class 05:50:38 -!- Pthing has joined. 06:02:41 zzo38: no, add all complex abilities 06:02:55 invocations, spells, powers, maneuvers, infusions, and anything else you can think of 06:03:06 except you randomly get certain kinds 06:03:22 so you never have access to all of them 06:04:23 hmm... I want a blank white cards bot 06:04:34 I don't think I will put the "randomly get certain kinds" but I could figure it out if (and only if) I can assign reasonable C.A.P costs to each of them, of course, you have a limited number of complex abilities known and a limited number of C.A.P/day so you can't possibly use all of them in one day 06:06:07 I don't know if I really want to add all of them though, if some kinds are more powerful it might require a feat to provide access to them 06:06:13 I was kidding 06:06:28 then again, I don't do much else with 3.5 06:06:35 except add templates to random creatures 06:07:03 * coppro still contemplates purchasing GR's book that's 100% templates, just for the lulz 06:07:45 -!- oerjan has quit ("Good night"). 06:09:10 OK 06:19:08 zzo38: ever played 1000 blank white cards? 06:19:17 No, I never have done so 06:19:29 But I have seen descriptions and rules 06:24:19 sounds retarded 06:25:09 it's awesome 06:26:34 hi!!! 06:27:00 a tad late, but otherwise well played 06:36:04 "Brad Cox and Tom Love create Objective-C, announcing "this language has all the memory safety of C combined with all the blazing speed of Smalltalk." Modern historians suspect the two were dyslexic." 06:36:17 I may have pasted that in here verbatim before 06:38:10 I haven't seen it before 06:38:13 and I lold 06:38:44