00:23 -!- tacco\unfoog has quit [] 10:19 -!- FIQ has quit [Ping timeout: 256 seconds] 10:25 -!- FIQ has joined #tnnt 13:45 @Tone @Luxidream you guys around? 13:49 <@Tone> Yes 13:49 <@Tone> Although I'm working 13:51 well we cant have that 13:51 :P 13:51 when you have time, do you mind going back over what your concerns were with scoring for tnnt? 13:53 <@Tone> Sure 13:55 <@Tone> IIRC it was mostly that scoring seemed to emphasize conducts, and in particular last year added a bunch of "simple" conducts that would end up being worth a lot of points since each one was multiplicative 13:56 <@Tone> I can give some examples later though 13:57 <@Tone> but even doing a bunch of the 0.1 conducts would add a lot of score 13:57 <@Tone> So it felt like there was some level of conduct/score bloat going on there 13:58 ok 13:59 yeah just whenever you have time to go into details. we're about to start work on this years tournament, the main emphasis this year is tweaking/correcting things with the scoreboard. no new game content this year. maybe some new achievments but that'll be up to aosdict 13:59 <@Tone> I think my personal impression after the tournament was that it might work better keep all the conducts and make them worth a good number of points but only count them once per tournament, or something like that 13:59 <@Tone> There are a few other ways that you could go about it though if you wanted to change something 14:00 <@Tone> but yeah I'll discuss more later 14:00 cool 14:03 contrasting that, we want someone to be able to do an illit and then a foodless illit and get credit for ascending two games with difficult conducts 14:07 <@Tone> Yeah one of the other ideas I remember having was to also award points for certain combinations of conducts, like foodless atheist or wishless polypile-less 14:08 <@Tone> Would probably require manually making all the combos which may not work well and also would possibly leave you with a large list 14:09 <@Tone> Could make every unique combo count once but if there are a lot of relatively simple conducts you could swap in and out like petless or permadeaf that you can swap in and out to make the combo unique then it isn't as effective 14:10 <@Tone> Could possibly give diminishing returns for repeated conducts 14:11 <@Tone> or maybe increase the relative value of the base ascension to devalue conducts slightly across the board 14:13 <@Tone> Maybe converting some of the tnnt conducts (like ascending without killing the Wizard) to a trophy or something that otherwise you'd only get awarded points for once would help 14:13 <@Tone> So you could still repeat and combo stuff like foodless and illit but not milk those for more points 14:13 -!- Mandevil has joined #tnnt 14:13 -!- mode/#tnnt [+o Mandevil] by ChanServ 14:17 Was there any sentiment that conduct multipliers are too high even if you only consider the 14 vanilla ones? 14:23 like, a few easy conducts plus a medium difficulty one with a role it meshes with = double the points of an asc 14:23 <@Tone> That's something I'll have to take a closer look at later, I'm just going on memory from last year right now :) 16:58 -!- oh6 has joined #tnnt 17:14 -!- tacco\unfoog has joined #tnnt 17:27 -!- amateurhour has joined #tnnt 18:35 <@luxidream> I'm alive 18:37 :partypopper: 18:37 no you're not alive you're just an AI outputting text on my screen 18:40 heh 18:40 sup luxi 18:46 <@luxidream> hey K2 18:47 hows things 18:49 -!- oh6 has quit [Ping timeout: 272 seconds] 18:50 <@luxidream> doin ok, still staying in my house but what can ya do :p 18:52 luxidream: are you a bad enough dude to help fix tnnt scoring balance? 18:53 <@luxidream> I'm looking through the tnnt score calculator right now 18:55 <@luxidream> I think the more general issue is that conducts don't have any kind of diminishing returns, so that makes it easy to find some optimal combination and farm them 19:00 <@luxidream> even if you remove all TNNT specific conducts, the multipliers for foodless and possibly zen could be used to get double the points of a normal ascension 19:02 apply z-score to All Things™ 19:02 hah 19:05 <@luxidream> I think that would work, apply z-score to conducts with full points where the conduct would have highest value 19:18 that smells suspiciously like an np-complete optimization problem 19:23 <@luxidream> something that the DCSS tournament does better is that turncount/realtime/streak length receive points based on your best, rather than fixed points for all of them, which encourages going for high achievement over farming 19:23 hmm 19:24 <@luxidream> Currently Tone has pointed out that streak points are too low considering you probably should break conducts to keep a long streak going, and I don't think turncount points are worth much for non maud players 19:24 so, we should encourage more elite playing? 19:25 under this system, if I'm getting you right, if you have 10 regular ascensions and a foodless, and someone else has 1 foodless, you're equal in the eyes of the tournament except you might have a most ascensions trophy 19:28 <@luxidream> You still receive points for each ascension of course, so no, that wouldn't be the case 19:31 <@luxidream> In the DCSS tournament it's still possible to receive bonus points with average scores across all three categories 19:31 <@luxidream> So it doesn't exclude more casual players 19:33 <@luxidream> A better example would be if one player has 2 10k turn ascensions and another has a single 5k turn ascension, the second player should receive more bonus points 19:37 <@luxidream> re: conducts it would probably just be easier to assign full value to conducts in the ascension with the highest overall multiplier 19:38 that's the optimization problem 19:38 if you move the z-scored-at-100% conduct from one asc to another, does it mean that's no longer the highest overall multiplier for other conducts, and so on 19:38 this might be interesting to tackle mathematically 19:39 homework: reduce this problem to 3-SAT :P 19:55 -!- hothraxxa has joined #tnnt 19:55 dammit i thought i was lurking in here. now i missed everything :( 19:56 Same here hothraxxa 19:57 https://www.hardfought.org/tnnt/irclog/%23tnnt-2020-09-16.log 19:59 -!- K2 changed the topic on #tnnt to: The November Nethack Tournament 2020, coming soon to a terminal near you :: https://www.hardfought.org/tnnt :: public irc logs available at https://www.hardfought.org/tnnt/irclog 20:00 oh showing us up now are you? :P 20:03 :P 20:04 the disclaimer about public irc logs should have been added to the topic a long time ago 20:04 * K2 shrugs 20:04 big brother is always watching u 20:09 i usually do my own logging so i never remember that there are public logs 20:15 -!- elenmirie_ has joined #tnnt 21:07 -!- aoei has joined #tnnt 21:12 -!- oh6 has joined #tnnt 21:36 -!- oh6 has quit [Quit: boopboop boop] 22:22 -!- amateurhour has quit [Ping timeout: 258 seconds] 22:57 -!- Tangles has quit [Ping timeout: 265 seconds] 23:11 -!- tacco\unfoog has quit [] 23:43 <@Tone> aosdict: It doesn't look quite as bad with vanilla conducts only and using last year's scoring system 23:46 <@Tone> I think the ascensions I did last year that were worth ~500 points would be worth ~120 or so 23:46 <@Tone> compared to a base ascension value of 50 23:47 <@Tone> I'm not counting Elberethless or Bonesless in there, not sure what constitutes "vanilla" for this 23:48 <@Tone> Weaponless Vegan monks would be worth a decent points boost 23:48 <@Tone> but even if you did those plus, let's say, genocideless, artiwishless, polyselfless, polyobjectless; you are looking at ~100 points 23:49 <@Tone> foodless is the only high value one that is probably the most exploitable 23:50 <@Tone> you can scum for archon starts to do pacifist but I can't imagine many people doing that repeatedly because of the nature of the conduct 23:51 <@Tone> and you still wouldn't break 150-200 points unless you added some conducts that would make it difficult 23:51 <@Tone> I guess someone with a start scumming script could scum a wizard start with =oSD and a smoky potion wish for an archon 23:52 <@Tone> not very likely though 23:52 <@Tone> So reducing it to vanilla-only conducts seems reasonable 23:53 <@Tone> also the discussion you all had earlier with luxidream had some good opinions in it 23:53 <@Tone> diminishing returns could work very well 23:53 <@Tone> would be interesting to see a system like that applied to last year's scores