Tournament Scoring: A Third Variant

Like the current VEKN system, except as follows: 1) A whole victory point is granted for last man standing, withdrawal, and survival to time out (ie... whenever not ousted). 2) For tournament ranking, both game wins and victory points are dispensed with. In its place, a victory score is used, based on victory points as follows: Victory Victory Points Score 0 0 1 1 2 10 3 100 4 107 5 115 3) Tournament points are used for 2d tier tie breakers; there is no third tier. ------------------ As long as there are less than 10 preliminary rounds, The hundred points for 3 vp is the functionally similar to a game win trumping any number of victory points under the current system -- except that 3 vp is effectively treated as the new game win requirement.

[ quoted text not captured ] Please explain the rationale for your values for Victory Scores. There seems to be logic behind them, but I think anyone would want to know that in order to properly dissect your proposal. Jeff

[ quoted text not captured ] Why should your fourth VP be worth less than your second? For one so concerned with zero-sum and distortion, you're making awfully strange propositions. Also, if you want to go with this, it would be better for 1 VP to be worth 2 points. You see a problem, where there is none. Even if this was a problem your solution would be worse than this hypothetical problem. Cheers, Bram

Jeff Kuta wrote: > Please explain the rationale for your values for Victory Scores. > There seems to be logic behind them, but I think anyone would want > to know that in order to properly dissect your proposal. Like all my proposals, this seeks to restore the original vp system, and to award time-out scores consistent with the original paradigm. My first and second versions sought to preserve -- fairly closely -- the current conditions for a "game win". However, this meant the definition for a game win had to be changed, since the elimination of half vps resulted in players with a single oust in time out games gaining 2 vps, and therefore scoring game wins provided no-one else had done the same (as when a 5-player table times out with 4 players remaining). To prevent this, but otherwise preserve game wins conditions, I presented the following definition: "3 vps are required in time-out games; whereas in completed games one needs more vp than any other player." XZealot and others complained that this definition was "too clunky". On later consideration, I found no reason not to change the definition, for tournament scoring purposes, to a much more simple & elegant "3vp = game win". The 2-vp game win, besides being rather rare, did not seem to me to be the sort of game win that was particularly deserving of high honors. Moreover, it seemed to me that I was adequately loyal to the spirit of the original rules provided the 2 vp "game winner" received a higher score than anyone else at his particular table (which he always does). He did not need to be ranked as high as a 3 vp game winner, nor need he be ranked any lower than the two players who (to use Salem's phrase) "tie for the win" at another table. Having thought of making 3 vp consistently equal a "game win", I found no reason not to merge game wins into the "victory score" idea that I had introduced in my second proposal. A game win trumps any lesser score (or any variation among higher scores). Since there are rarely (if ever) more than 3 preliminary rounds, this can be effectively simulated by having the score for a 3- vps win be more than 3 times higher than any lesser score (or any variation among higher scores). In this case, I used a factor of 10 (just because it seemed convenient). Similarly, to provide maximum "risk incentive" to a player who has not bothered to score a vp yet, I create a case where a single two vp score trumps any number (such as 3) of 1 vp scores. Once again, I use a factor of 10. No attempt is made to provide risk incentives to players who have already scored two ousts. A player with 2 ousts has far more to lose by being ousted than he has to gain by ousting. This is no different from the current VEKN system (where no vp bribe can outweigh the game win the player has to lose), and no different from my earlier proposals. Nor is any attempt made to provide a risk incentive to the player with 3 ousts scored and one player left to devour. I really don't think this guy needs any fire under his butt. As it happens, a slight risk incentive is provided (7 to lose versus 8 to gain), which is rather less than the current VEKN system (.5 to lose versus 1.5 to gain). I could made 4 and 5 worth 110 and 120 respectively, but decided to avoid tie scores. Hence, 4,0,0 is worth less than 3,2,0 but more than 3,1,1. 5,0,0 is worth more than 3,2,0, but less than 3,2,2.

Bram Vink wrote: > Why should your fourth VP be worth less than your second? It isn't really, in terms of practical effect. For instance, I could make the first vp worth 3 points instead of one, and then (by your standards) the fourth would be worth the exact same thing as the second. But this would have no meaningful effect on outcomes in a tournament with no more than 3 preliminary rounds. It might be more accurate to say that the fourth vp is worth more than the first vp, but less than the first 2 vps combined. > For one so concerned with zero-sum and distortion, you're making > awfully strange propositions. The zero-sum-ness of the vp system itself is preserved. > Also, if you want to go with this, it would be better for 1 VP to > be worth 2 points. 3 points would be more consistent with your objection, but I do not see how it could make any practical difference. > You see a problem, where there is none. > Even if this was a problem your solution would be worse than this > hypothetical problem. Explain how it would cause problems.

[ quoted text not captured ] ... and the sliding scale of VP scoring. i.e., instead of a flat, say, +10 VS per VP, you've got 1 for the first, 9 for the second, 10* for the third, 7 for the fourth, and 8 for the fifth. * Assuming 80 VS for the GW. Not sure what is being accomplished with that variance. That is, I don't see why the fourth is worth less than the third and less than the fifth. Or why the first VP is worth almost nothing. One interesting change in the system vs. the current is the devaluation of consistency vs. feast-or-famine that the TP system provides. If player A has a 3-2, 2-3, 0-5 finish over three tables and player B has a 3-2, 1-4, 1-4 finish, the current system ranks B over A (on tournament points), while the proposed system ranks player A higher (since VP #2 in the second game is worth more than VP #1 in the third). And, to be sure everyone follows along with the "current system" parallel alluded to above, here's the current system (or rather, a functionally-equivalent recasting of the current system): VPs VS --- --- 0 0 1 10 2< 20 2> 120 3 130 4 140 5 150 Where 2< indicates 2VPs but no GW (i.e., someone else has 2 or more VPs in that game). And 2> means 2VPs with a GW. You can recast the per-VP score to any suitably small number, of course. The 10 used was just an example. 1 would also get the same result: VPs VS --- --- 0 0 1 1 2< 2 2> 102 3 103 4 104 5 105

In message <1193651750.4...@19g2000hsx.googlegroups.com>, nys...@cs.com writes: >Bram Vink wrote: >> Why should your fourth VP be worth less than your second? > >It isn't really, in terms of practical effect. Sure it is. > For instance, I could >make the first vp worth 3 points instead of one, and then (by your >standards) the fourth would be worth the exact same thing as the >second. But this would have no meaningful effect on outcomes in a >tournament with no more than 3 preliminary rounds. In a 2 round tournament, I get a game win and 3 VP and come second with 2 VP. Total TP: 108 (60+48). Another player gets 4VP on one table (and a game win) and comes second on another table with 1 VP (where another player got the other 3 or 4). Total TP: 108. 1 GW, 5 VP, 108 TP - tie. In your point system, I get 110VS - 100 (3VP) + 10 (2VP). In your point system, the other guy gets 108VS - 107 (4VP) + 1 (1VP). Result: 3+2 beats 4+1, no tie. Why is 3+2 better than 4+1? -- James Coupe PGP Key: 0x5D623D5D YOU ARE IN ERROR. EBD690ECD7A1FB457CA2 NO-ONE IS SCREAMING. 13D7E668C3695D623D5D THANK YOU FOR YOUR COOPERATION.

[ quoted text not captured ] So a 3 VP GW is an order of magnitude greater than 2 VPs. Which is fundamentally the same as the current system. But what is the rationale behind 1,10,100, 107,115 trending? Why isn't 4 VPs worth 1000 points and 5 VPs worth 10,000 points? You seem to indicated that each VP is an order of magnitude greater than the preceding VP. This is inelegant and needs to be explained in a manner that is transparent. Comment Welcome, Norman S. Brown, ,Jr XZealot Archon of the Swamp

In article <1193693968.5...@d55g2000hsg.googlegroups.com>, XZealot <xze...@cox.net> wrote: > So a 3 VP GW is an order of magnitude greater than 2 VPs. Which is > fundamentally the same as the current system. > > But what is the rationale behind 1,10,100, 107,115 trending? Why isn't > 4 VPs worth 1000 points and 5 VPs worth 10,000 points? You seem to > indicated that each VP is an order of magnitude greater than the > preceding VP. I can't speak for the author, but I am guessing it is: -Make 1VP worth something, but far less than a "GW". -Make 2VP worth something, but far less than a "GW". -Make a 3VP "GW" the main goal of the game. -Give you a bonus for 4 and 5 VPs, but a small one that makes them more or less equivalent to single VPs, but cuts down on the tie situations that would show up if all the VPs were equal. Given a system of: 1VP=1 2VP=10 3VP=100 4VP=107 5VP=115 you have the most incentive to get a 3VP "GW", and any extra VPs over that are just minimal tie breaker points. But if the system were: 1VP=10 2VP=20 3VP=100 4VP=110 5VP=120 Ties would show up all the time. And with the proposed system, getting a 5VP sweep is worth more than 2VP in game A and 3VP in game B. Which may or may not be something that is something that is necessary. But ok. That being said, I'm unclear why this is a better system than the one we currently have. Peter D Bakija pd...@lightlink.com http://www.lightlink.com/pdb6/vtes.html "Find hungry samurai." -The Old Man

James Coupe wrote: > In message <1193651750.4...@19g2000hsx.googlegroups.com>, > nys...@cs.com writes: > >Bram Vink wrote: > >> Why should your fourth VP be worth less than your second? > > > >It isn't really, in terms of practical effect. > > Sure it is. > > > For instance, I could > >make the first vp worth 3 points instead of one, and then (by your > >standards) the fourth would be worth the exact same thing as the > >second. But this would have no meaningful effect on outcomes in a > >tournament with no more than 3 preliminary rounds. > > In a 2 round tournament, I get a game win and 3 VP and come second > with 2 VP. Total TP: 108 (60+48). Another player gets 4VP on one > table (and a game win) and comes second on another table with 1 VP > (where another player got the other 3 or 4). Total TP: 108. > > 1 GW, 5 VP, 108 TP - tie. > > In your point system, I get 110VS - 100 (3VP) + 10 (2VP). > In your point system, the other guy gets 108VS - 107 (4VP) + 1 > (1VP). > > Result: 3+2 beats 4+1, no tie. > > Why is 3+2 better than 4+1? Good point. That result is incongruous. If anything, it would be more consistent with the general theme of the proposal to have the extreme scores beat the more-average scores. If anything, 4,1 should beat 3,2, just as 3,0,0 beats 2,2,2 and 2,0,0 beats 1,1,1. This can only be achieved by an escalation in intervals value (ignoring the whopping 90-point "game win" interval at the 3 vp level). Perhaps the following would work better: Victory Victory Point Score 0 0 1 3 2 10 3 100 4 108 5 117 However, I am unsure about whether 3,5 should beat 4,4, or whether that gives too much advantage to 5-player tables. Perhaps the escalation should not continue after 4, and the value should be 116, or even pull back to 115. One could create a situation where the pattern was reversed for extremes centered on 3,3 (ie. 3,3 beats 4,2 which in turn beats 5,1) but holds true in all other cases: Victory Victory Point Score 0 0 1 3 2 10 3 100 4 109 5 115 Or one could make it balanced but skewed at the extremes: Victory Victory Point Score 0 0 1 3 2 10 3 100 4 107 5 110 This tends to promote ties, which I'm not sure is good, now that half vps are eliminated. Or elimiante "extra" risk incentives and make it completely balanced: Victory Victory Point Score 0 0 1 1 2 2 3 100 4 101 5 102 This also tends to promote tie scores.

LSJ wrote: > ... and the sliding scale of VP scoring. > i.e., instead of a flat, say, +10 VS per VP, you've got 1 for the first, 9 for > the second, 10* for the third, 7 for the fourth, and 8 for the fifth. > > * Assuming 80 VS for the GW. > > Not sure what is being accomplished with that variance. > That is, I don't see why the fourth is worth less than the third > and less than the fifth. Well, I have decided that it was a mistake to make the second worth more than the fourth, as indicated in my post to Mr. Coupe. I am also not sure it is not a mistake to have the fifth worth more than the fourth, since this might give 5-player tables too much of an advantage. Part of my motivation is to avoid ties. I am a bit worried that by eliminating half victory points, I make tie scores too common. However, I don't want to do this arbitrarily. There should be some logic it. > Or why the first VP is worth almost > nothing. Well, that's just maximum risk incentive -- giving players (almost) nothing to lose compared to time out, and everything to gain. Nothing slows a game down like 5 players playing cautiously and conservatively, so if any risk incentive is needed besides the game win itself, the place to focus that incentive is on those 5 starting players who have not scored ousts yet. Once players start getting knocked out, things start to speed up naturally. Any value for 1 vp that was less than 1/3 of the value for 2 vp would have achieved the desired effect (in a tournament with no more than 3 preliminary rounds). I liked the "propaganda value" of giving a score of 1 instead of 3. I added this change shortly before posting without adequately thinking through other consequences. I'm still not convinced that any risk incentive is needed besides the game win.