An Ambiguity in the Blind Auction System

[NoPunIn10Did]

Secondary question: do the rules allow for the following:

Player 1 bids 5 points on Country A and has it in 5th place, with two other countries on lower points.

Player 2 bids 5 points on Country A and has it in 7th; last place.

Each player has no other country selection on 5 points.

Is it a tie break or does Player 1 get Country A by virtue of the higher ranking? I suggest the rules do, and should, grant a tie a break because we are not just trying to avoid least favourites, and these two players like Country A the same amount.

That should be a tie break, yeah--cause the first thing looked at is points allocated, with rank being used as a tie breaker, not as the primary determining factor.

Player 1 would end up with Country A though, because rank is the first tiebreaker, and they had it ranked higher.

I think this might be a good example of why unique integer bids may be worth the extra math. If I was that player who bid 5 on my seventh place and didn’t get at least a coin flip against the other player’s 5, I’d feel cheated. It might be the rules, but the optics are pretty terrible.

Another related optics issue is that of having so many 1’s on the auction block. It isn’t actually the case that loads of players are getting one of their least-favorite picks, but it sure can look that way.

With unique integer bidding, the number is the only value that matters. A 5 is a 5. A 1 is a 1. Order is irrelevant, and it’s considerably easier to automate in a spreadsheet.

[Nanook]

Secondary question: do the rules allow for the following:

Player 1 bids 5 points on Country A and has it in 5th place, with two other countries on lower points.

Player 2 bids 5 points on Country A and has it in 7th; last place.

Each player has no other country selection on 5 points.

Is it a tie break or does Player 1 get Country A by virtue of the higher ranking? I suggest the rules do, and should, grant a tie a break because we are not just trying to avoid least favourites, and these two players like Country A the same amount.

That should be a tie break, yeah--cause the first thing looked at is points allocated, with rank being used as a tie breaker, not as the primary determining factor.

Player 1 would end up with Country A though, because rank is the first tiebreaker, and they had it ranked higher.

I think this might be a good example of why unique integer bids may be worth the extra math. If I was that player who bid 5 on my seventh place and didn’t get at least a coin flip against the other player’s 5, I’d feel cheated. It might be the rules, but the optics are pretty terrible.

Another related optics issue is that of having so many 1’s on the auction block. It isn’t actually the case that loads of players are getting one of their least-favorite picks, but it sure can look that way.

With unique integer bidding, the number is the only value that matters. A 5 is a 5. A 1 is a 1. Order is irrelevant, and it’s considerably easier to automate in a spreadsheet.

Why would they have a right to feel jipped? If they wanted it, they should have bid more points on it or ranked it higher.

Let's take a scenario of Player 1 bidding 30 points on Country A as his 4th place bid, while player 2 bids 30 points on Country A as their highest bid. If that goes to a coinflip and P2 loses, how is that fair to him?


Now, if Player 1 had bid 31 points, that would be a different matter. But yeah, I think rank as the first tiebreaker is very fair...and I'm confused why this is controversial at all.

[NoPunIn10Did]

I think this might be a good example of why unique integer bids may be worth the extra math. If I was that player who bid 5 on my seventh place and didn’t get at least a coin flip against the other player’s 5, I’d feel cheated. It might be the rules, but the optics are pretty terrible.

Another related optics issue is that of having so many 1’s on the auction block. It isn’t actually the case that loads of players are getting one of their least-favorite picks, but it sure can look that way.

With unique integer bidding, the number is the only value that matters. A 5 is a 5. A 1 is a 1. Order is irrelevant, and it’s considerably easier to automate in a spreadsheet.

Why would they have a right to feel jipped? If they wanted it, they should have bid more points on it or ranked it higher.

Let's take a scenario of Player 1 bidding 30 points on Country A as his 4th place bid, while player 2 bids 30 points on Country A as their highest bid. If that goes to a coinflip and P2 loses, how is that fair to him?


Now, if Player 1 had bid 31 points, that would be a different matter. But yeah, I think rank as the first tiebreaker is very fair...and I'm confused why this is controversial at all.

It’s “controversial” from my standpoint because using number value AND order to judge across multiple players is an easily misunderstood rule, no matter how clearly you think you’ve described it. Either the explanation is ambiguous and short, or it’s well-explained but verbose.

As such, it’s the sort of thing where:
1) a player can easily misunderstand what their bid’s actual value is compared to others’.
2) the GM won’t be able to catch that misunderstanding beforehand.

Requiring unique integer bids provides an equivalent system that takes less text to explain and where the most likely misunderstandings (equal bids or bids that add up above the max) can be caught (by the GM) and corrected (by the player) prior to the auction.

[Woolgie]

Imagine a bank manager has a series of appointments, they are every 10 minutes and each lasts for 10 minutes. The first is booked for 9am.

Now imagine the bank manager starts work only at 9.10 am. There are two possible options:

1. Everyone is seen 10 minutes late and no one gets their preferred time.

2. The person at the front of the queue is moved to the back and then every other person does get their preferred appointment time but that one person gets a really rubbish time.

Which is better?

In the world of diplomacy the analogy is: do you give as many people as possible their preferred country choice and fit the rest to any old country, or why not abandon the idea of anyone having to get their first choice and instead calculate awarded countries based on giving the best all-round deal? If everyone got their second choice rather than some getting their first choice and some getting their third choice, would that be so bad?

Invariably someone ends up with their fifth, sixth, seventh choice just so most people can have their first choice. But if I we're that person with the 9am appointment I'd be pretty annoyed.

[NoPunIn10Did]

Imagine a bank manager has a series of appointments, they are every 10 minutes and each lasts for 10 minutes. The first is booked for 9am.

Now imagine the bank manager starts work only at 9.10 am. There are two possible options:

1. Everyone is seen 10 minutes late and no one gets their preferred time.

2. The person at the front of the queue is moved to the back and then every other person does get their preferred appointment time but that one person gets a really rubbish time.

Which is better?

In the world of diplomacy the analogy is: do you give as many people as possible their preferred country choice and fit the rest to any old country, or why not abandon the idea of anyone having to get their first choice and instead calculate awarded countries based on giving the best all-round deal? If everyone got their second choice rather than some getting their first choice and some getting their third choice, would that be so bad?

Invariably someone ends up with their fifth, sixth, seventh choice just so most people can have their first choice. But if I we're that person with the 9am appointment I'd be pretty annoyed.

Are you suggesting we abandon the existing algorithm and move to one that optimizes for highest overall consensus? There are ways to do that, but calculating consensus by hand can get really ugly (and explaining that algorithm concisely is a bear).

It also creates scenarios where somebody's lower-end bids can hurt their chances at getting their upper-end bids.

Say for instance that nobody wanted to play Italy. If Players A, B, C, D, E, & F put Italy as their lowest bid, while Player G put Italy as their second-lowest bid, then there are a number of scenarios where Player G might be assigned Italy prior to their highest bids being evaluated at all.

This is what I call the "nice guys finish last" effect. No matter what algorithm we use, someone will be disappointed, but there's a pretty stark difference in the disappointment of "I lost the coin toss," and that of "I didn't get a coin toss even though I bid the max points."

[Woolgie]

Touché, NoPun, it is technically possible that the Italy scenario you describe could happen but I wonder how often? I’d like to run some simulations and get back to you.

To be clear, I do prefer this so called ‘least overall rank’ method, probably using a Hungarian Algorithm. When I played Dip by email this was the method used. The layman’s explanation is, each country is ranked 1 to 7 by each player and the goal of the GM is to award the least overall rank. If everyone gets their first pick then the overall rank is 7 points. If everyone gets their fourth pick then the overall is 28. Six players getting their first pick and one player their seventh pick would give an overall rank of 13.

[Nanook]

Imagine a bank manager has a series of appointments, they are every 10 minutes and each lasts for 10 minutes. The first is booked for 9am.

Now imagine the bank manager starts work only at 9.10 am. There are two possible options:

1. Everyone is seen 10 minutes late and no one gets their preferred time.

2. The person at the front of the queue is moved to the back and then every other person does get their preferred appointment time but that one person gets a really rubbish time.

Which is better?

In the world of diplomacy the analogy is: do you give as many people as possible their preferred country choice and fit the rest to any old country, or why not abandon the idea of anyone having to get their first choice and instead calculate awarded countries based on giving the best all-round deal? If everyone got their second choice rather than some getting their first choice and some getting their third choice, would that be so bad?

Invariably someone ends up with their fifth, sixth, seventh choice just so most people can have their first choice. But if I we're that person with the 9am appointment I'd be pretty annoyed.

Now you're getting into a preference based system as opposed to a points based system. This is also a pretty terrible analogy for an auction-style, points based system, cause the people with appointments didn't bid on them, they got them on a FCFS or a preference basis.

[NoPunIn10Did]

Touché, NoPun, it is technically possible that the Italy scenario you describe could happen but I wonder how often? I’d like to run some simulations and get back to you.

To be clear, I do prefer this so called ‘least overall rank’ method, probably using a Hungarian Algorithm. When I played Dip by email this was the method used. The layman’s explanation is, each country is ranked 1 to 7 by each player and the goal of the GM is to award the least overall rank. If everyone gets their first pick then the overall rank is 7 points. If everyone gets their fourth pick then the overall is 28. Six players getting their first pick and one player their seventh pick would give an overall rank of 13.

Even if you modify the example I mentioned such that Italy becomes that player’s second choice, there still remains this problem:

Assume P is the number of players (and also the number of powers being bid on).

In the standard algorithm, if you make your top bid with the maximum points, you can guarantee that at minimum you’ll have a 1/P chance at receiving your first choice regardless of what anyone else bids. Likewise, in a purely random assignments system, you’re guaranteed a 1/P chance as well.

In a consensus system, though the overall result is better in aggregate, a player’s max bid provides no such guarantee.

[NoPunIn10Did]

First stab at revised rules are now posted and open for comments/editing:

viewtopic.php?f=34&t=57868&p=949820#p949819

[EpicDim]

Coming late to this discussion, but as a mathematician I have always processed it as a coin flip because 1 is the highest bid remaining and since you are ignoring everything all the non-earliest order equal bids, Germany for Phil and Germany for Quentin are the only two bids currently in the matrix. So it's a coin-flip.

I don't think you should take overall position into account at all. It should be position based on what's left. It's just luck that with only 2 left, it's my 3rd preference and your 4th preference. Just because my 4th preference got bid on higher by someone else. But overall with only two remaining it's both of our first pick. That's what the original algorithm intended to do.

And even though some people say that "everyone can use a spreadsheet", I can tell you from experience that asking 7 people (even when they were all software developers) to give me preference lists with no duplicates and that add up to X. I got 5 correct. I personally like the original algorithm and if GMs want to use it, then great. But if you're going to get rid or duplicate numbers, you HAVE to do it for the no preference as well.