An Ambiguity in the Blind Auction System

[Woolgie]

I’d argue the benefit of simplicity in using B without any points allocation, just rank. That’s how I always played dip before this site.

[NoPunIn10Did]

I’d argue the benefit of simplicity in using B without any points allocation, just rank. That’s how I always played dip before this site.

That's an okay system too, but the point-value auction gives more flexibility for a player to express their preferences. Take these two players, for example:

Code: Select all

Typical Joe         No Italy Irene
50  Germany         21  Germany
35  France          20  France
 5  Russia          19  Russia
 4  Austria         18  Austria
 3  England         17  England
 2  Turkey           4  Turkey
 1  Italy            1  Italy

In an order-only system, these two players would be represented as having the same exact bids. In both cases, Italy is the player's last choice, but Irene is able to express that she doesn't really care which of her top five she gets, but that she'd rather not have Turkey or Italy (especially not Italy).

Now, if everyone bids the exact same set of numbers (as with the hypothetically optimized case), then it devolves into an order-only system anyway. That's not necessarily a problem.

But if even one player wants to express a difference of opinion for how they actually feel about their picks, a points-based system lets them do so.

[VGhost]

It should be B...his overall pick was higher...but I agree most GMs probably do it wrong.

I think that both B and C are doing it wrong, though. Knemeyer's rules don't specify how the rankings of same value bids are to be judged across players (rank overall or rank-within-value), in all likelihood because those rankings aren't supposed to be judged across players.

This is why I specified that it should be publicly stated.

There are really two questions here:

1) How should Knemeyer's rules be resolved?
2) Do additions to those rules add any benefit?

I think you're right that there's an ambiguity; but I would rather add a rule if that keeps situations from getting to a coin-flip.

[NoPunIn10Did]

There are really two questions here:

1) How should Knemeyer's rules be resolved?
2) Do additions to those rules add any benefit?

I think you're right that there's an ambiguity; but I would rather add a rule if that keeps situations from getting to a coin-flip.

Knemeyer's existing rules should be resolved as version A, with order of same-valued bids only mattering within one player's own bids, and not compared against other players.

While my stated Option #3 preserves Knemeyer's rules as-is, just making them clearer, my actual preference would be Option #1 with unique integer bids. That system reduces the possibility of coin flips respective to Knemeyer's rules.

Now if you really want to get rid of more coin flips beyond that, you'd have to implement something along the lines of what sinnyb suggests (a consensus-building system that suspends higher tied bids until lower untied bids are resolved first), but that has other ramifications and can create a sort of "nice guys finish last" effect not present in the Knemeyer algorithm or any of these suggested modifications.

[Nanook]

Option B I see a lot, and option B is what I generally use. Does anyone here use option C? I can't recall seeing it in any game I've been in...do you have an actual example of someone here using C instead of B?

[NoPunIn10Did]

Option B I see a lot, and option B is what I generally use. Does anyone here use option C? I can't recall seeing it in any game I've been in...do you have an actual example of someone here using C instead of B?

In this particular instance, it didn't matter, but you did.

All the same-value bids for a player are marked with their rank-among-equal-bids, not their rank-overall, so unless you had something else planned...

I'm not meaning to pick on you here, to be clear. Aeschines did the same thing in Greek City-States, though by happenstance it didn't end up mattering there either.

[joe92]

I was tripped up on this when I first GM'ed ModEx. I ran it with the understanding in my mind that it was meant to be A. It created a large uproar and I was forced to redo the bidding process. I debated during that redo that I think it should be A. I said I was going to bring this up for debate but I didn't as since then I've lessened my opinion and am actually more in favour with B being the process. I see a merit to B in that you can go for a maximum bid for 1 power and if you don't get it, then the 1 point bids are ordered according to preference meaning the chances of getting your absolute last choice are minimal. Since we all agree that the stated aim of the bidding process is to achieve exactly that - minimum chance of getting least favourite power - then option B is better than option A. Option C is just a slight alteration to option B which actually complicates it a little bit more than it needs to be.

The idea for asking for unique point bids for each character is good on one level, but bad on another. It makes the bidding process more difficult for the player. If you have a 14 power map and the total bidding points is 210 with a max of 140 on 1 bid, then the arithmetic around calculating the bids for the other powers is harder than some people are comfortable with. It shouldn't be necessary that a player needs to grab a calculator to get their bids in. As a GM it doesn't affect me, I've got a spreadsheet doing it all for me, but the players don't. Since this is just the same as option B with simply the benefit of making it clearer by demanding unique bids I don't think the extra difficulty is worth it.

Personally I think that the rules we all link to from Pedros (well, his direct quote of Knemeyer) are outdated and confusing. I have wanted someone to rewrite them for a while. I cringe every time I link to that thread. It does not clarify the process very well.

I think that the rules should be kept as are. I think that the rule as we interpret it regarding the order of the 1 point bids (or any bid) should be kept but there should be a new version of the rules written out that clarifies everything in the simplest and easiest of terms to understand. It should be unequivocal when reading the rules exactly what the procedure is. I don't support the idea of requesting unique point bids. While most players here are more than capable with that level of arithmetic I don't see the benefit of adding that extra difficulty when it needn't be there. The same effect can be reached with equal bids and attributing importance to the order.

Also, what Knemeyer meant or didn't mean is pretty much irrelevant. He wrote out draft 1 and continued up to draft 10 of a set of rules for bidding. There's an unlimited version of new drafts and improvements to be written for those rules.

[Nanook]

Option B I see a lot, and option B is what I generally use. Does anyone here use option C? I can't recall seeing it in any game I've been in...do you have an actual example of someone here using C instead of B?

In this particular instance, it didn't matter, but you did.

All the same-value bids for a player are marked with their rank-among-equal-bids, not their rank-overall, so unless you had something else planned...

I'm not meaning to pick on you here, to be clear. Aeschines did the same thing in Greek City-States, though by happenstance it didn't end up mattering there either.

The notation indicates which order the same number bids were in, but the actual adjudication was according to points allocated>rank, with the notation only meant to mark rank within an individual player's bids.A player with their third one bid as their 6th ranked bid overall would've beat your one point last rank bid, for example.

To be fair though, I can see how the notation was confusing.

[NoPunIn10Did]

Since we all agree that the stated aim of the bidding process is to achieve exactly that - minimum chance of getting least favourite power - then option B is better than option A.

Minor quibble: the goal of the bidding process isn't simply to provide "minimum" chance of getting least favorite power. What we're trying to handle here is a version of the Stable Marriage Problem, the solutions to which can vary based on what we prioritize. If we really wanted to prioritize not-getting-last-pick as the highest goal, that would require a vastly different approach.

However, I do agree that it is one of the goals of the process, along with generally finding a fair way to resolve persons' first/second/etc. picks to the best of our ability too.

Option C is just a slight alteration to option B which actually complicates it a little bit more than it needs to be.

I see option C being used periodically, but probably not intentionally. I'm not sure if all GM's have even realized that Options C and B make a difference, and there are plenty of games where they don't. While Option B can still be translated into a system where each player still gets an equal "budget" of points to work with, Option C can actually yield more points to some players than others.

As such, no matter what path we take, I think we should ditch Option C entirely.

The idea for asking for unique point bids for each character is good on one level, but bad on another. It makes the bidding process more difficult for the player. If you have a 14 power map and the total bidding points is 210 with a max of 140 on 1 bid, then the arithmetic around calculating the bids for the other powers is harder than some people are comfortable with. It shouldn't be necessary that a player needs to grab a calculator to get their bids in. As a GM it doesn't affect me, I've got a spreadsheet doing it all for me, but the players don't. Since this is just the same as option B with simply the benefit of making it clearer by demanding unique bids I don't think the extra difficulty is worth it.

I think this is a fair criticism, particularly when dealing with larger games. I still prefer having unique integers, but I can understand that not being the default. Unique integers do tend to handle a lot better in spreadsheets that automatically calculate the bid result, though I gather many folks just process bids by hand anyway.

If we are going to use option B as our official interpretation moving forward, I'd also like there to be clear rules on that, along with some guidelines about how to display those order values in a manner that's consistent and easily comparable in a table of bids.

Personally I think that the rules we all link to from Pedros (well, his direct quote of Knemeyer) are outdated and confusing. I have wanted someone to rewrite them for a while. I cringe every time I link to that thread. It does not clarify the process very well.

Agreed. That's what I was hoping this thread would prompt: consensus on how to move forward. If there's consensus here, I'd be happy to do the first draft; after editing we'd make a permanent post of it (and link to it accordingly).

I'll make another reply here with a short proposal.

[NoPunIn10Did]

To be fair though, I can see how the notation was confusing.

If we keep option B as the default for PbF, I'd like us to come up with a more standardized notation. If someone have 5 1-point bids out of 7 total picks, then I think it would be less ambiguous to do one of these:

1. List overall preference rank on every bid, even if that player didn't submit multiple bids of that value (since they might still need to be cross-compared).
   
   • 50 (1)
   • 45 (2)
   • 1 (3)
   • 1 (4)
   • 1 (5)
   • 1 (6)
   • 1 (7)
2. Convert preference rank into a numeric value (with top rank being a value that looks like the player count).
   
   • 50.07
   • 45.06
   • 1.05
   • 1.04
   • 1.03
   • 1.02
   • 1.01

Option ii would be my personal preference, as that appeals to me from an automation standpoint, but I don't know if that would confuse folks more.