Thursday, April 06, 2006

The Problem with Blind Bidding

"1 ..."

I had two, perhaps three seconds left to make a decision, and it would decide the entire game.

"2 ..."

We'd tied at rock last time, and the average person shifts upward, which meant Eric was most likely to go to paper. But Eric was bright, and he probably knew that, which meant he'd stay at rock to crush my scissors when I displayed that to cut his paper if he did shift up as expected, but if ...


My gyrating hand came to stuttering stop, stuck on rock because I was frozen in indecision.

And Eric had counted on my indecisiveness, as was evidenced by by his wide spread hand.

"Paper smothers rock!"

To be honest, I've never understood why paper beats rock in the first place.

Stupid game.

If there's a single game mechanic that's more maligned than randomness (for which see my articles The Problem with Luck and The Problem with Luck II), it's blind bidding: simultaneously making selections at the same time as other players.

As usual, I think that the nay-sayers don't understand the mechanic.

However, I think that some designers don't understand the mechanic either, because I've played games with good blind bidding mechanics, and I've played games where the blinding bidding mechanic is bad. To my frustration, bad blind bidding designers don't seem to understand the problems with their game, and thus I can to some extent empathize with those players who have been burned.

A Definition of Terms

I've been using the term "blind bidding" thus far, because it's the term most frequently used when people explain how blind bidding games fall somewhere between Monopoly and Candyland in their strategic basis. However, I don't think it's really the correct term to describe what's actually a broad class of very similar, but subtly differentiated mechanics.

To better define that class of games I'd use the phrase simultaneous selection (though that term is sometimes used exclusively for the second subcategory I note below). Broadly, there are three major subcategories of simultaneous selection.

First is what I actually consider to be blind bidding. This is the the simultaneous selection of differently valued markers. It's an auction mechanic, what's commonly called a closed-fist or sealed-envelope auction. You can find it in Fist of Dragonstones and Modern Art.

Second is simultaneous action. This is simultaneous selection of different actions; often these actions become less valued when more people take them. This mechanic is found most obviously in Basari and Goldbrau. I think it's the best classification for Rock, Scissors, Paper too.

Third is simultaneous ordering. This is the simultaneous selection of what order a set number of items are evaluated in. This has been used as the combat system for both Age of Mythology and Dungeonville (though not successfully in either case).

As usual, I've got lots of reviews of these games if you'd like some more insight:

Blind Bidding Games: Caribbean (B+), Fist of Dragonstones (B-), For Sale (A), Modern Art (A-), Money! (B), O Zoo Le Mio! (B-)

Simultaneous Action Games: Basari (B), Goldbrau (C+), Hoity Toity (B), Rock, Scissors, Paper (D-)

Simultaneous Ordering Games: Age of Mythology (C+), Dungeonville (C), Fairy Tale (A)

Some of these games do seem to cross the boundaries.

With Caribbean it looks and feels like bidding, but on the other hand you get your money back each turn, and you have to bid in very discrete amounts; it would have been easy to classify it as "simultaneous ordering" instead.

Meanwhile Fairy Tale is just barely simultaneous ordering, but the order can matter infrequently (with "hunt" and "flip" cards). If it weren't for that, I could probably have classified it as simultaneous action, since each turn you select 3 actions from a set of 5.

But no definition's perfect ...

The Problem with Design

One of the reasons that blind bidding had gotten a bad rap is because some people don't understand how to design it well. Roshambo (or Rock, Scissors, Paper) offers an example of gameplay that just barely works, and sadly there's worse than that in actual published games.

In order for any type of simultaneous selection to be meaningful, the different choices have to have different values. In Roshambo this is marginally true because of the selection method. Players have no way to make a truly random selection among the three options while playing in person, and so human psychology tends to influence the result. Players will start off with rock more often than the other two options, and as I noted above, people will often move upward, choosing the option that would have won the last round.

If Roshambo were played with a method where people could randomize their choices, say if each player had a rock card, a scissors card, and a paper card, and they could either select one or truly randomize one, then any strategy in the "game" would disappear.

The wikipedia page on Rock, Scissors, Paper has a fine explanation on this strategy devolution:
Mathematically optimal play (according to game theory) is a simple matter of selecting randomly, and so the game may be considered trivial in that sense when played in a way that eliminates psychology, as with a computer. But "optimal" in this sense means only "incapable of being defeated more than expected by chance", while it does not imply that the random strategy is best at taking advantage of a suboptimal opponent. In fact, if the opponent is human or a non-random program, it is almost certain that he plays suboptimally and that a modified strategy can exploit that weakness.
Now the problem with some simultaneous selection game design is that: (1) like Roshambo there's no true difference between the value of the selections; and (2) you often have cards which you can randomize to make a selection. If you're the player that's playing suboptimally, you should immediately switch to randomizing your selections, thus giving a superoptimal player who could read your moves only a few marginal wins before play returns to the random chance at the heart of this gameplay.

Dungeonville and Age of Mythology are two published games which fall into this exact trap. In each game you're setting up one-on-one battles between yourself and an opponent, and you're doing so by ordering your troops so that your first troop will fight your opponent's first, your second will fight his second, ... etc. In Dungeonville in each battle the better (lower numbered) troop wins. In Age of Mythology each troops tends to have a advantage against one other type of troop and a disadvantage against another, and that will often decide the result.

To be clear: This. Mechanic. Doesn't. Work.

It amounts to an entirely random selection unless one player is stupid, and pretending anything else just tends to waste time, wherein players carefully try to select an order that really doesn't matter.

And it gives a bad name to blind bidding, simultaneous selection, simultaneous ordering, or whatever else you want to call it.

The Problem with Play

Now a well designed simultaneous selection does something very different: it ensures that different options have different values. As a result, a random selection is no longer superior to a non-random selection, because you might be getting high-value results less often than you would have if you'd actually selected.

Pure blind bidding does this sort of by default, because if you bid more for something, that's clearly more cost to you (thus making the item you're bidding for lower value), and if you bid a little, that's lower cost. It's simultaneous action and simultaneous ordering games that have to think about this design more carefully.

If you examine the design of Basari, it's a fine example of game that does place different values on the different options. First of all, any of the options (points, gems, movement) may have a higher value to you personally. You might currently have a high-point value or a high-gem value depending on what space you're on. Second of all, each of the options has an overall desirability which might increase or decrease its personal worth to you depending on how many people you expect to take it. If everyone is on a high-point value space, you might feel the value of gems is higher to you, because of the lower likelyhood that anyone else will take it.

The heart of a good simultaneous selection design is, thus, the ability to make guesses about which optimal actions other people will take ... and then start second-guessing whether they'll actually take those optimal actions or not.

Which brings us to the heart of what a simultaneous action system should mean to a player, and it's a term that I've talked about before: risk vs. reward.

In Basari and any other good simultaenous selection game, there's usually a clear best choice to you, but taking that clear best choice is often risky, because there's opportunity for other people to spoil it. Thus, you decide whether you want the best reward, and if you do, you understand you're taking a risk to try and attain that reward.

And, just like in those situations with controllable luck that I've been talking about lately, if you risked too much and you lost ... it's your fault. Period. End of story. Quit complain'.


Simultaneous selection games--including blind bidding, simultaneous action, and simultaneous ordering--should have at their heart the same basis as any luck-based game: the ability to assess risk versus reward.

If a game doesn't, because all options are equally valued and thus the various simultaenously selected rewards are the same, then the simultaneous selection mechanic has probably been badly designed. You were right. The blind bidding sucks.

But, most simultaneous selection games do have the ability to make that assessment. If you want to win them--and you can, and you can do so with some reliability--then you need to undestand that you're taking your risk if you're going for the big reward.

And live with it.

By happestance last night I played three of the games listed herein: two games of Fairy Tale, one game of Beowulf, and one game of Money!. I did well in the Beowulf--despite having really bad luck with wounds--but winning at Fairy Tale and Money! remains beyond me. In this three games I picked up two fifth-place results and one third.

That I could do so reliably poorly in the two pure simultaneous selection games (Fairy Tale and Money!) just points out the strategy implicit in this sort of game--and apparently, my inability to grasp it.


GROGnads said...

To begin with, then it is "mechanism" while WITH a "non-working mechanic", then HE is probably some "Dealership" sort of 'fellow'! I also take exception upon those idiotic "News"-folk who inform US all of "suspicious" packages, instead OF "suspect" kinds. Just WHAT are all of them 'packages' being "suspicious" upon, you may wonder? Why, probably about whether or NOT that they soon will "blow UP real GOOD!" due to some over-reacting "numb-wit" being 'a-skare`d with such! And yeah, I'm taking OUT my "frustrations" upon one and ALL! so WATCH 'it'!

Anonymous said...

Another simultaneous selection is Phantoms of the Ice/Slapshot/Team. The first time through the deck, we shuffle, but in the sudden death OT, it's a lot of fun. It does have the variable value that you mentioned as a criteria, though. Adel Verpflichtet is another where the values are worth differing amounts and another that I enjoy a lot.

Pawnstar said...

[aside]I'm with grognads; garage mechanic yes, game mechanic no.[/aside]

One game I enjoy that employs this simultaneous selection is Raj/Hol's der Geier. It's short and quick; the entire game is simultaneous selection.

I also like the way it works in GoldbraĆ¼. As I recall it gets easier to second-guess your opponents as the game wears on (as some choices are clearly detrimental); a lot like what you were saying about Basari (never had the pleasure - played ES&R once).

Chris Farrell said...

A game clearly has "mechanics" (check it out on But "mechanics" is wierd in that it can be singular or plural based on construction. So it's not clear you can singularize mechanics when it's not clear whether it's plural or not.

I prefer to use the words "system" or "systems", myself. But my feeling is that "mechanic" is a legitimate usage, more legitimate than a lot of the strange vocabulary we use in describing games. And it passes the test of "everyone is perfectly clear on what you mean when you say it" when established terms don't really exist.