Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: 1 2 3 [4] 5  All

Author Topic: I don't know what this thread has turned into (was Making it to Level 42 -Lvl 5)  (Read 23877 times)

0 Members and 1 Guest are viewing this topic.

Kirian

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 7092
  • Shuffle iT Username: Kirian
  • An Unbalanced Equation
  • Respect: +9371
    • View Profile
0

(S)He does use automatch. Someone in this topic stated that they know tat uses automatch because they got automatched against tat. Some people who automatch don't accept direct challenges because... you know... they automatch and a direct challenge is not an automatch. Try automatching when tat is around. See if you get automatched against him (her). If (s)he declines that match, THEN you can come back here and whine about tat. Until then you can't.

S/he may use the automatch function, but s/he does not automatch in spirit--or, put a different way, s/he is gaming the automatch system.  When I'm automatching, out of every 10 games I usually hit 2 or 3 players at my level +-10 (15-35).  Tat doesn't seem to play anyone above level 15 or so, ever.  Which means s/he is declining any automatch games against higher-level players.

See, playing automatch is supposed to make things sort of a melange of Swiss, random, and round robin.  In a tournament setting, not showing up for a match means you forfeit.  On the ladder, however, declining a match does nothing to your rating.  So you can jump in and out of automatch status, declining anyone with a rating higher than 15, and it doesn't affect you.

While TINAS is technically correct, the chances of someone reading this thread, who is above level 15, getting randomly matched with tat, is pretty low.  And if he doesn't accept match proposals, then obviously it's not something we can directly test unless this random happenstance occurs.

But then, maybe he does accept proposals and not only automatches... in which case someone just needs to propose to him from a registered level 1 or 2 account...
Logged
Kirian's Law of f.DS jokes:  Any sufficiently unexplained joke is indistinguishable from serious conversation.

Blooki

  • Bishop
  • ****
  • Offline Offline
  • Posts: 118
  • I am constantly overmeditated...
  • Respect: 0
    • View Profile
0

FWIW, I've been automatched with tat and declined.

For full disclosure though, I will decline matches with top players if I'm just not in the mood for an intense game.

I have no problem with tat's selectivity with opponents. I would think that one day he/she would want to see how he/she stacks up against some other top players.
Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3594
  • Respect: +6035
    • View Profile
    • Dominion Strategy
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.  It is much, much more difficult than it seems to consistently beat randomly-selected bad players.

Also, whatever tat's doing, it's certainly more ethical than Larry's route of hitting #1, which was playing fake opponents.
Logged

timchen

  • Minion
  • *****
  • Offline Offline
  • Posts: 704
  • Shuffle iT Username: allfail
  • Respect: +233
    • View Profile
0

It seems lots of people care about the ranking. So... why don't we have a more serious one? Since there is already this auto-match function, I don't think it requires a lot of modification. Rules may be something like this:

(i) you don't get to choose your opponent. The system chooses one player randomly, with a higher probability toward one with similar levels: (say a Gaussian distribution, with standard deviation 10 levels) once selected you cannot decline.
(ii) you can only play the same opponent, say 2 times a day.
(iii) you may ban say up to 5 cards, but you may not select cards.
(iv) 3 seconds after every move, there is a clock counting. For each move there is a max, say 3 mins to think, and 10 mins total for a game. (Mainly exist to prevent from dragging opponents, since you cannot decline them)

Does this sound fair?
Logged

ackack

  • Explorer
  • *****
  • Offline Offline
  • Posts: 302
  • Respect: +19
    • View Profile
0

- I think a serious ladder should not allow any card bans at all, or if it does use them they should be universal.

- Some games are inherently lengthy, and action chains can take a long time to play out. I think this makes timing games difficult. I think you want something more like the Fischer clock - a set amount of time for the whole game that is incremented each turn. You'd then keep the individual turn time out, as well.
Logged

Kirian

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 7092
  • Shuffle iT Username: Kirian
  • An Unbalanced Equation
  • Respect: +9371
    • View Profile
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.

Agreed.  You need to have a 75%+ record against low-level players to get much benefit out of it.  I'm sure we could calculate the approximate ratio.  Tat does better than that, which suggests that playing against high-level opponents would, likely, be just as good if not better for him/her.  But then again those matches might take longer.

It seems lots of people care about the ranking. So... why don't we have a more serious one? Since there is already this auto-match function, I don't think it requires a lot of modification. Rules may be something like this:

Heh, your first two rules are pretty much the matching rules for Starcraft 2.  Decline an automatch?  You forfeited.  I find the addition of a timer interesting, but I wouldn't set a final time-control like that; certain boards are going to take longer than 10 minutes per side.
Logged
Kirian's Law of f.DS jokes:  Any sufficiently unexplained joke is indistinguishable from serious conversation.

timchen

  • Minion
  • *****
  • Offline Offline
  • Posts: 704
  • Shuffle iT Username: allfail
  • Respect: +233
    • View Profile
0

It seems I didn't state the timing rule I have in mind clear enough. You have (say) 3 seconds per move (play an action, buy a card, etc); only after the 3 seconds the timer starts to tick. This way I think 10-minute total thinking time is fair. 
Logged

rspeer

  • Witch
  • *****
  • Offline Offline
  • Posts: 469
  • Respect: +875
    • View Profile
0

It seems lots of people care about the ranking. So... why don't we have a more serious one? Since there is already this auto-match function, I don't think it requires a lot of modification. Rules may be something like this:

(i) you don't get to choose your opponent. The system chooses one player randomly, with a higher probability toward one with similar levels: (say a Gaussian distribution, with standard deviation 10 levels) once selected you cannot decline.
(ii) you can only play the same opponent, say 2 times a day.
(iii) you may ban say up to 5 cards, but you may not select cards.
(iv) 3 seconds after every move, there is a clock counting. For each move there is a max, say 3 mins to think, and 10 mins total for a game. (Mainly exist to prevent from dragging opponents, since you cannot decline them)

Does this sound fair?

TrueSkill is designed for automatching, and it always surprises me that DougZ doesn't use that part of it (I suppose it has fiddly edge cases when few people are automatching, and it might not be worth the time to code it when the official version might drop any week). The idea is just to choose players so that their win probability against each other is close to 50%.

Banning up to 5 cards sounds quite exploitable. You could ban all the attacks that dilute your deck the most, for example, and then you basically only have to know how to play money-centric strategies.

I like the idea of a timer.

Of course, all this would depend on either Isotropic logging more information (who declined which games, who played when, etc.) that could be used by a separate ladder, or the mysterious developers of the official version listening to these ideas.
« Last Edit: July 28, 2011, 01:02:10 pm by rspeer »
Logged

Captain_Frisk

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1257
  • Respect: +1261
    • View Profile
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.  It is much, much more difficult than it seems to consistently beat randomly-selected bad players.

I will second this.  I'm guilty of caring about my rating too much, but the advent of automatch has killed me.  I've been as high as 42... and now i'm stuck down at 36.  Previously I always challenged the highest ranked player. 

I think the "problem" is that the win probability between levels can vary quite a bit, and varies significantly based on the cards on the table. 

Some tables have very obvious or limited strategies.  Those games are likely to be "flatter" in terms of the relationship between win% and player skill.  Other games might have cards with significant variance. 

Using treasure map as an example.  Without help, treasure map is usually a bad play.  If I played 100 games with theory, and he went un-assisted treasure map every time, I would expect to beat him about 60-70% of the time.  However, when it hits on turn 5 - its going to be very tough to beat.  I don't remember the exact % of the time, but it is very far from zero. 

Lets assume that it's 30%.  There's the "good" strategy... that theory and tat know... and there's the "buy only treasure map strategy", that anyone can see.  If the true skill model looks at theories level, and then a lvl 5 guy (we'll call him "chump", and assumes that the Chump only has a 12% chance to win against the great and all powerful theory, then all Chump has to do is play treasure map every time, and he'll gain levels while theory loses them. 

So - by putting some control over the table, you can influence your ranking.  Ideally - if you're a good player, you want to bias the table to cards that reward skillful play with minimal variance.  If you're a bad player, you want to increase the variance as much as possible, to make the game result much closer to a coin flip.

You can look at the logs of folks out there to see what constraints they put on their games... and come to your own conclusions. 

One final leaderboard gaming tip:

Because isotropic assigns starting player advantage based on most recent result, any time you lose, you should make sure you use that advantage on as high ranked opponent as you can.  The first player advantage is anywhere from 2-15% (its 8% for tat, and 12% for theory), and you want to cash that in for as much as you can.

I agree with most of the comments that the leaderboard should be fully random (seat order chosen randomly, you don't know who your opponent is, what cards are chosen, no constraints).  Maybe this should be a separate leaderboard, but as long as you can control your the shape of your game and your opponents, there will be people who will game it.  I'm half tempted to intentionally throw 50 games in a row to someone just to skyrocket them to the top.

Then, you gotta remember that its just a number, and of the N thousand people playing dominion on isotropic, there are probably only 50 really care about the topic.


Logged
I support funsockets.... taking as much time as they need to get it right.

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 987
  • Uncivilized Barbarian of Statistics
  • Respect: +1177
    • View Profile
    • CouncilRoom
0

I dislike

(iii) you may ban say up to 5 cards, but you may not select cards.

but otherwise like your scheme.  Alternatively, you may ban cards, but then your rating is penalized to some degree for playing not quite uniformly random dominion.
Logged

DG

  • Governor
  • *****
  • Offline Offline
  • Posts: 4071
  • Respect: +2611
    • View Profile
0

I had a look back a number of Tat's games in the council room and they were well played, in my humble opinion. Most of the losses came from bad draws and/or heavy attacks and there were very few blunders. I hope nobody ever looks back through my games since there are some truly awful games as well as the good ones. Tat's style seemed to be based on creating very strong engines that cashed in late for multiple vp cards each turn. It would be nice to see that tested against opponents who can acquire vp much faster and don't allow him/her to get 9 hunting parties.

As for selecting games for ranking points, well I guess I'm failing at that. I mostly just let other people challenge me.
Logged

fp

  • Thief
  • ****
  • Offline Offline
  • Posts: 94
  • Respect: +5
    • View Profile
0

Is this "player bashing" really necessary?

How do we know Larry was playing fake opponents? How do you know tat is intentionally playing weaker players in order to inflate his or her rating? We don't!

Maybe players are gaming the system. But who cares? It is only a silly leaderboard, and a simple fact about ranking systems (i.e. leaderboards) is that they are never perfect. There is always going to be a way to game them; there is always going to be glitches. That is just how it works.

EVEN IF players are gaming the system in shady ways, we shouldn't be "bashing" players of things we "think" or "suspect" they might be doing. They might log on here (or might already be) and feel very unwelcome very quickly.

Heck, they might just be doing it to see if they can do it.

If anything, we should be suggesting ways to fix the system as opposed to berating players that might appear to gaming it.

Speaking of the leaderboard, where is theory?
Logged

Taqman

  • Herbalist
  • **
  • Offline Offline
  • Posts: 8
  • Respect: 0
    • View Profile
0


Speaking of the leaderboard, where is theory?

Taking the bar exam I expect.  I believe your name disappears if you're inactive for about 10 days? 
Logged

Blooki

  • Bishop
  • ****
  • Offline Offline
  • Posts: 118
  • I am constantly overmeditated...
  • Respect: 0
    • View Profile
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.  It is much, much more difficult than it seems to consistently beat randomly-selected bad players.

I will second this.  I'm guilty of caring about my rating too much, but the advent of automatch has killed me.  I've been as high as 42... and now i'm stuck down at 36.  Previously I always challenged the highest ranked player. 

Curious... because I am of the inclination that I would drop precipitously in ranking if I didn't automatch and challenged the highest rank player. Weird.

Maybe it's the sheer volume of my games, maybe there's a bug in the code that ranks me differently, but my personal experience seems to indicate that losing to low-level players isn't a big deal. I quite frequently lose games to low-level players unnecessarily for non-game-related reasons (I started a game I didn't have time to finish, my internet connection can be spotty and I get disconnected somewhat often, I'm on raging-tilt and rage resign on the first few turns because I have no patience for even the slightest things like boards I just don't feel like playing ... OK maybe this last one is game-related). I do recall dropping 30-40 rankings after a particular rage-induced streak, but was soon back in the Top 10 and I don't recall any notable victories over top players during that comeback.

I don't really know what I'm trying to say other than I believe my experience playing low level players is drastically different from that of others. I think I'll try challenging the highest ranked player for a period and see how it goes.
Logged

ackack

  • Explorer
  • *****
  • Offline Offline
  • Posts: 302
  • Respect: +19
    • View Profile
0


Speaking of the leaderboard, where is theory?

Taking the bar exam I expect.  I believe your name disappears if you're inactive for about 10 days? 

I thought it was 2, actually. Not very many, at any rate.
Logged

theParty

  • Chancellor
  • ***
  • Offline Offline
  • Posts: 20
  • Respect: +2
    • View Profile
0

Wow - I just got sucked into reading this whole thread.  I had no idea there was this kind of vitriol around the top of the leaderboard.

I'm nowhere near the top of the leaderboard (high 20s, low 30s lately), but I think the real reason to care about your rank (at least why I care about my rank) is that you can get matched against better players.  I suppose some people want to play lower ranked opponents to beat up on them, but a lot of people (myself), want to play similar or better opponents for a good game, or to get better.  Auto-matching has obviously changed things, but I decline auto-matches all the time from lower ranked players.  They may be good, and they may beat me if we played, but it's not a very fun game when I am auto-matched against someone and they buy a mine instead of a mountebank with their 5/2 opening.

That's why (I think) your rank matters, and why I care about mine.  I have no idea how the ranking system work (beyond, win and you move up proportionately to who you beat), but it's pretty accurate in that people towards the top of the leaderboard usually beat me, and I usually beat people who are 10 levels below me.

Logged

Agrisios

  • Chancellor
  • ***
  • Offline Offline
  • Posts: 24
  • Respect: 0
    • View Profile
0

I believe your name disappears if you're inactive for about 10 days? 
I thought it was 2, actually. Not very many, at any rate.

I think it's a week.
Logged

painted_cow

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 151
  • Respect: +19
    • View Profile
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.

Agreed.  You need to have a 75%+ record against low-level players to get much benefit out of it.  I'm sure we could calculate the approximate ratio.  Tat does better than that, which suggests that playing against high-level opponents would, likely, be just as good if not better for him/her.  But then again those matches might take longer.

It seems lots of people care about the ranking. So... why don't we have a more serious one? Since there is already this auto-match function, I don't think it requires a lot of modification. Rules may be something like this:

75 % wins against weak opponents isnt that hard in 1on1. I got 72 % in 3 player games with automatch (many weak players, but sometimes at least a level 20+ player...). With a lesser amount of variance in 2 player games reaching that numbers is not that much of a deal, if you play thoroughly and with reason.

And guys, its the leaderboard...I cant imagine of anythink more important when playing, since there are like no Tournaments out there.
Logged

Thisisnotasmile

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1493
  • Respect: +672
    • View Profile
0

I don't want to get into this discussion too much, but if anyone thinks that playing bad players is a good way to boost your level, I would encourage you to try it.

Agreed.  You need to have a 75%+ record against low-level players to get much benefit out of it.  I'm sure we could calculate the approximate ratio.  Tat does better than that, which suggests that playing against high-level opponents would, likely, be just as good if not better for him/her.  But then again those matches might take longer.

It seems lots of people care about the ranking. So... why don't we have a more serious one? Since there is already this auto-match function, I don't think it requires a lot of modification. Rules may be something like this:

75 % wins against weak opponents isnt that hard in 1on1. I got 72 % in 3 player games with automatch (many weak players, but sometimes at least a level 20+ player...). With a lesser amount of variance in 2 player games reaching that numbers is not that much of a deal, if you play thoroughly and with reason.

And guys, its the leaderboard...I cant imagine of anythink more important when playing, since there are like no Tournaments out there.

Fun?
Logged

kn1tt3r

  • Minion
  • *****
  • Offline Offline
  • Posts: 585
  • Respect: +278
    • View Profile
0

I mean who really cares about the exact position in the leaderboard? Basically players around 20 are good, but not great. Above 30 players really know what they do and are hard to beat... well, and >40 are just top-notch and the best around.

Anything beyond this differentiation is just childish and doesn't get you anywhere.
« Last Edit: July 30, 2011, 09:57:43 am by kn1tt3r »
Logged

a714generation

  • Pawn
  • **
  • Offline Offline
  • Posts: 1
  • Respect: 0
    • View Profile
0

I mean who really cares about the exact position in the leaderboard? Basically players around 20 are good, but not great. Above 30 players really know what they do and are hard to beat... well, and >40 are just top-notch and the best around.

Anything beyond this differentiation is just childish and doesn't get you anywhere.

I think different people get different levels of interest/enjoyment out of the true skill system, but I tend to subscribe to kn1tt3r's point of view. I like to know what general percentile of players I'm surrounded by, but whether Blooki or Tat is the best player doesn't really matter. Others feel differently, and that's cool.

Playing against Blooki or DG, or Theory, or Yaron, or whoever is in that top range of players is always going to be a fun challenge. The important thing about this site is that we're building a community of players of different skill ranges who can challenge each other, formulate new strategies, and most importantly, have fun.
Logged

Captain_Frisk

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1257
  • Respect: +1261
    • View Profile
0

I mean who really cares about the exact position in the leaderboard? Basically players around 20 are good, but not great. Above 30 players really know what they do and are hard to beat... well, and >40 are just top-notch and the best around.

Anything beyond this differentiation is just childish and doesn't get you anywhere.

Agreed... we should all calm down.  But even you point out that people with level >40 are top notch and the best around.  Right now there are 15 people who meet that criteria, and another ~50 within striking distance.  The question is... does the current ranking system really accurately rank those people, or are there ways to game the system?  If there are ways to game the system, should it be improved to prevent that "abuse"?

Selfishly, I'd like a way to reliably compete against other top players.  Some folks, like theory & Celicath are pretty good about accepting any challenges (the fact that theory finally ended up with a winning record against me upsets me to no end).  Other folks, I've never been able to play.

For the remaining 5,400 people on the leaderboard, the system works pretty well.  Any change made for the 100 people who really really care about their rank shouldn't come at a burden for the remaining 98%.
Logged
I support funsockets.... taking as much time as they need to get it right.

fp

  • Thief
  • ****
  • Offline Offline
  • Posts: 94
  • Respect: +5
    • View Profile
0

I think it's a week.

In any case, we will find out. I am off for a week.
Logged

DG

  • Governor
  • *****
  • Offline Offline
  • Posts: 4071
  • Respect: +2611
    • View Profile
0

Metzgerism is looking at the format and options for the bgg league again. If anyone wants a competitive league they can contribute some time and opinions. http://www.boardgamegeek.com/thread/592845/commissioners-take-on-the-bggdl/page/6
Logged

Jimmmmm

  • Torturer
  • *****
  • Offline Offline
  • Posts: 1734
  • Shuffle iT Username: Jimmmmm
  • Respect: +1944
    • View Profile
0

Hi all...
Just wanted to announce my existence by saying that I just beat tat!  ;D
We played very similar games, utilising Shanty Town, and not too many other action cards, and I managed to come out on top 29 to 28 after 15 turns each.

I am a lowly level 13 (although I imagine that has now been boosted somewhat) having played about 145 games on Isotropic and won roughly half.

Just to point out, he didn't use automatch, he proposed a game with me, prohibited Alchemist, Possession and Treasure Map, and gave us identical starting hands.

Anyway, I'm sorry if this comes across as if I'm talking myself up, I realise I'm not even close to being in the same league as a level 30+, I was just reading this thread the other day and thought I could contribute having played him.
Logged
Pages: 1 2 3 [4] 5  All
 

Page created in 0.12 seconds with 21 queries.