[Dom3] Brainstorming for a late era draft game.

[DrPraetor]

I am slightly paralyzed with indecision. Since it seems we have 5 people filling out the form for this test I need to fill in 5 commanders, scouts, priests gods, and misc commanders, 10 capital only troops and mages, 15 recruit anywhere mages and spells, 25 troops, and any national advantage I care to bid on?

Yes, that's right.

That is far too much to consider (almost 100 choices when backups are taken into account; even if I only put a 1pt placeholder bid on all my backups, it is still ~8% of my total), and a game would likely involve more people participating. Also, is there an order the bids are evaluated in? I either missed that or didn't understand it, so clarification would be nice.

Highest first. I'll give an example below.

Also I'm not sure that I like that people who bid nothing are averaged in. With the example of the warlock, if I decide that I can't make them work in my faction, or that I don't need them, I am not actually making a statement about the value of the unit for the guy that gets them. It seems like anyone with unorthodox choices will have a huge advantage under that system.

Well, people with unorthodox choices will get them.
But as Frank was complaining above - it doesn't do you any good to have points left over. So if you make a bunch of unorthodox choices, you'll get your unorthodox choices even if you also asked for a few powerful dominion advantages.
The people who made orthodox choices will either get their orthodox choices and no dominion advantage, or they'll get dominion advantages and have to take 2nd or 3rd pick on their orthodox selections.
I think that's fair; in any case, that's not a bug, it's a feature.

I'm also not particularly keen on an inability to adjust my choices if I miss a key unit. Say I go heavy for jaguar warriors and lay down 25% of my resources on him, but someone else gets 30% of their resources on them somehow and actually gets the unit. If I have been bidding on the assumption that I am supporting jaguar warriors, I am now kinda screwed, I have a bless god and probably some summons and mages that go with a strategy I can no longer run because I don't have the unit it hinged on. Moreover I am probably locked more fully into that strategy as all my points are now making other high bids stronger, so I almost certainly get everything I wanted to support jags, but no jags to support.

I agree that this is potentially a problem. I can put a favorable spin on it and say, "well, you're discouraged from depending on a particular choice" or something like that.
One option would be to do a follow-up. Whoever has points left over (starting with whoever has the most points) can then spend them to trade stuff they actually got for other stuff that wasn't drafted by anybody, to fix errors at the end. I think a single round of such fixing wouldn't be too time consuming.

Finally, a side issue, but whoever has jags will also have jag PD, which I think is kinda screwy. This applies equally to other high quality scary units, there is almost a certainty that someone will have skinshifter PD, and that someone will have Elephant PD, and Illithid PD.

Oy, you're right.

Fair enough - even though it opens the system up to be more prone to "gaming", your PD consists of your 4th and 5th unit selections, with your 2nd and 3rd unit selections coming at PD 20+.

If I got the explanation right it matters because if you got the Warlocks with 15% and in the next pick you and me both have an 10% at an pick I would get it because my 10% of X is bigger than your 10% of X-15%.

Right! Exactly! Simple, yeah? Counter-intuitive, but not complicated.

This makes no sense. If I bid 15% for Warlocks, what the fuck difference does it make how many points I spend? It's not like I get the 5% overbid back, because there's only one round of bidding.

From your description, everyone ends up with a variably sized pile of extra points and it doesn't matter in the slightest because you can't respend them.

Well, we could let people respend them in a single sweep at the end. That wouldn't be too burdensome. But, basically - if you make unorthodox choices, you'll get all your choices but not spend any points. I don't see how you can complain, though: you got all your choices! No system is completely idiot proof.

Consider a hypothetical situation, with 3 players, each of whom is going to draft a single troop and a single mage. The bids they turn in are:

unit/player	A	B	C
Warlock	1000	200	6
Theurg	0	200	3
Skratti	50	250	4
Sorcerer	0	0	5
Jotun Herse	0	150	1
Jaguar Warrior	500	100	3
Cavalry	1	0	0
Longbowman	1	100	1

So, first, the parser checks that everyone's bids are valid. A & C do not have valid bids. A did not bid on enough Mages (he needed to bid on 3 to guarantee he does get 1) so the parser fills in his last bid for him (this can be done at random but I think actually I'll put in a default priority system so that people can leave a lot of stuff blank if they wish, and not be hosed.) C, on the other hand, bid on too many mages. There's no way with only 3 players that he could possibly get his 4th choice, that's just trying to game the system. So his 4th choice is set to 0. This results in the following normalized bids, and the following costs:

unit/player	A	B	C	COST
Warlock	6439	2000	3000	3813
Theurg	6	2000	0	669
Skratti	322	2500	2000	1607
Sorcerer	0	0	2500	833
Jotun Herse	0	1500	500	667
Jaguar Warrior	3220	1000	1500	1907
Cavalry	6	0	0	2
Longbowman	6	1000	500	502

So, first thing that happens, the single highest bid in the table, which is A for the Warlock, is processed. Two things happen.
A spends 3,813 points, so his bids will now total to 10,000-3813 = 6187 points. Also, all the total base bids are reduced for each participant, as all of the now-unavailable bids are set to 0.

The updated table looks like this:

unit/player	A	B	C	COST
Warlock	3813	0	0	3813
Theurg	0	2500	0	669
Skratti	0	3125	2857	1607
Sorcerer	0	0	3571	833
Jotun Herse	0	1875	714	667
Jaguar Warrior	6162	1250	2143	1907
Cavalry	12	0	0	2
Longbowman	12	1250	714	502

The next highest pick in the entire table is player A's 6,162 bid for the Jaguar Warrior (I'll explain the moral in a bit). So player A gets the Jaguary Warrior, and his point total drops from 6,187 to 4,280. The bidding is over, and those 4,280 points can't be "spent" - but it hardly matters because he's got what he wanted already.

The new table looks like this:

unit/player	B	C	COST
Theurg	2857	0	669
Skratti	3571	3000	1607
Sorcerer	0	5000	833
Jotun Herse	2143	1000	667
Longbowman	1429	1000	502

It goes on, but long story short - C gets Sorcerer for 833 points, then C's bid for the Jotun Herse goes up to (10000-833)/2, and he gets the Jotun Herse (ties are broken toward the more expensive) for 667 points. Finally, B gets the Skratti for 1607 points and the Longbowmen for 502 points.

The moral of the story is that it is better to know what you want, and to go ahead and bid on it. With more than 2 rounds of bidding, the point costs will eventually start to add up to something meaningful, but the basic message isn't changed. Bid high one what you actually want. The Will to Power, that is the thing!

I'm considering, though, that it might be better not to unfreeze bid points until an option is taken by somebody - so that the people who bid high on one category don't get all their points freed up. It also might be a good idea to charge people their actual bid rather than the average? I'd like some people to make honest picks so that I can figure this stuff out.

[Username17]

So, first, the parser checks that everyone's bids are valid. A & C do not have valid bids. A did not bid on enough Mages (he needed to bid on 3 to guarantee he does get 1) so the parser fills in his last bid for him (this can be done at random but I think actually I'll put in a default priority system so that people can leave a lot of stuff blank if they wish, and not be hosed.) C, on the other hand, bid on too many mages. There's no way with only 3 players that he could possibly get his 4th choice, that's just trying to game the system. So his 4th choice is set to 0. This results in the following normalized bids, and the following costs

OK, that's just fucking insane. I am finding it difficult to understand how you could complete writing that without understanding that your idea was fundamentally shitty.

How about: you put a preference order to every single thing in each category? Everyone gets their first pick (roll offs for people who selected the same pick for number one), then everyone gets their second pick. You could either encourage people to fight for popular choices by having the losers of roll offs and people who selected something that has already been taken in a previous round get their next available choice that round, or you could discourage in-the-box thinking by having the player simply get nothing that round and just keep doing new rounds with everyone who doesn't have enough units until everyone has enough units.

I mean, it still has the problem where once I get fucked out of the Uba by some asshole getting it instead of me, then I end up with no death magic in my faction at all, which is kind of fucked. But hey, you could sell that as a feature - only a couple of factions would end up with full magic coverage.

But at least you wouldn't be doing complete bullshit where people are being forced to bid random or having their bids discounted and in any case being left with a giant (and variable) pile of points that they can't do a single thing with.

-Username17

[Korwin]

Sent the Excel.
It's definitly less fun than an draft. And no way to really plan some synergy. (will I get sacreds? If yes, which? Do I want an Pretender for an bless or not, etc.)

btw. DrPraetor you are up at the spell draft (in case you missed it).

[DrPraetor]

So, first, the parser checks that everyone's bids are valid. A & C do not have valid bids. A did not bid on enough Mages (he needed to bid on 3 to guarantee he does get 1) so the parser fills in his last bid for him (this can be done at random but I think actually I'll put in a default priority system so that people can leave a lot of stuff blank if they wish, and not be hosed.) C, on the other hand, bid on too many mages. There's no way with only 3 players that he could possibly get his 4th choice, that's just trying to game the system. So his 4th choice is set to 0. This results in the following normalized bids, and the following costs

OK, that's just fucking insane. I am finding it difficult to understand how you could complete writing that without understanding that your idea was fundamentally shitty.

How about: you put a preference order to every single thing in each category? Everyone gets their first pick (roll offs for people who selected the same pick for number one), then everyone gets their second pick. You could either encourage people to fight for popular choices by having the losers of roll offs and people who selected something that has already been taken in a previous round get their next available choice that round, or you could discourage in-the-box thinking by having the player simply get nothing that round and just keep doing new rounds with everyone who doesn't have enough units until everyone has enough units.

I mean, it still has the problem where once I get fucked out of the Uba by some asshole getting it instead of me, then I end up with no death magic in my faction at all, which is kind of fucked. But hey, you could sell that as a feature - only a couple of factions would end up with full magic coverage.

But at least you wouldn't be doing complete bullshit where people are being forced to bid random or having their bids discounted and in any case being left with a giant (and variable) pile of points that they can't do a single thing with.

-Username17

The idea with the points is to allow the market to price different features. We want that, remember, because it's the only way to fairly price the advantage of picking early for mages vs. the advantage of picking early for spells vs. the advantage of getting reanimating priests etc.? Having points left over at the end isn't really a problem - because the only purpose of the points is to break ties in terms of who gets to pick next. We would like the points at the end to be as even as possible, so: How about, instead - the person with the most points gets the single thing that they wanted most? When there is a tie at that stage, it goes to whoever has the single highest bid.

So going back to my example from above -
[*] Everyone has the same number of points to start with, so A has the highest bid so A gets Warlock.
[*] B and C are tied for number of points, B has the higher bid so B gets Skratti.
[*] C has the most points, gets Sorcerer. C still has the most points, gets Jaguar Warriors.
[*] B gets Jotun Herse.
[*] A gets Longbowmen.

Yeah, that's much better.

It still requires you to trim people's bids automagically.

[Username17]

Having points left over at the end isn't really a problem

Yes it is. If you have points left over that have no utility to you, then the market hasn't cleared and the free market has not done its job.

You're just doing preference voting combined with stupid shit.

Market pricing on the abilities only makes any sense if the players get the unspent points back and get to buy other things with them in the future. If there are points left in everyone's wallet at the end, the marginal utility of a point is zero and there's no market in any reasonable sense of the term. It's just weighted preference voting where the weighting is that if you win a game of odd or even with the other players then you get an extra advantage over the other players and they can go suck it.

-Username17

[DrPraetor]

Having points left over at the end isn't really a problem

You're just doing preference voting combined with stupid shit.

It's just weighted preference voting where the weighting is that if you win a game of odd or even with the other players then you get an extra advantage over the other players and they can go suck it.

No, but it is preference voting.

Suppose we have 5 players. A, B, C and D all bid high on a few items they really want, while E spreads his bids around.

Suppose A, B and C get their first choices, spending a bunch of points. Then, D and E take turns getting whatever they wanted. Then, A, B, and C take turns from among the rest.

At the end of this, D and E probably have more points than A, B or C. The marginal value of a point at the end is 0 - but the marginal value of those points during the bidding process is not 0, and it's still a market. It's just a market for oranges using teenage girls as currency (they expire if not used - we've had this conversation before!). Some people are going to leverage their oranges more effectively, but in the case above -
[*] A, B and C choose to take 1 or 2 things with high priority, and then had to choose later on as a consequence. They can hardly complain.
[*] D underbid on what he wanted. That sucks for him, but this is hardly an unfair method of denying him what he wants. He's going to end up with points at the end to reflect the fact that he was outbid, and his consolation was getting all of his second-tier picks serviced first.
[*] E either bid low on everything, or bid on a bunch of things no-one else wanted. Either way, E can't complain about the result, as E got what he wanted.

This seems like it works to me. Obsessing over the fucking accounting is irrelevant - suppose that I changed the rules so that things got more expensive as the rounds went on, such that everything was scaled and everyone spent exactly all their points (still always buying the single remaining item on which they bid highest). How exactly that would that be preferable? Are you saying that would be more of a market because everyone spent all their points - even though, because having a large number of points gives you a shorter run of consecutive selections, the marginal value of saving a point in the early rounds is actually lower?

What matters is the result. So either fill out the damn form with the points you think are fair, so I can run the social example and see if this really works, or stop bitching.

[Korwin]

While I did not enjoy the excel bidding, I do want to see what nation I would have gotten.

The motivation for the excel is to speed the process up.
How about to instead do an normal draft? But shorten the things you do an draft on?

Do we really care what non magic Commander every one has?
Hell I dont even care what priest I have, except when I am doing an bless and even then I would try to get an mage with H3 and continue to not care what non mage priest I have.

The same with name and colour, why an extra draft? Just fold it into another draft if you really want...

[DrPraetor]

Korwin - I think the way to handle this is to hand everyone a sheet with default (low) values for most of the draft.

This is what I'll do in the future. Then, you go into the sheet and you raise the #s for anything you care at all about, and that should be relatively quick. So people who really think tramplers are good can bid a bit on some tramplers, and so forth.

I agree that it's not as much fun as a draft, and you don't get synergy decisions, and so forth. So probably we'll want to a draft instead. But set up properly this method should be way faster - you think it took too long to fill out the sheet, okay. But the draft game in the other thread has been drafting stuff for three weeks. That's nothing on the scale of how long the game itself may last, of course .

[Korwin]

No, I don't think it took long to fill out the sheet.
It was unfun to fill out.
(The draft was/is fun. For me at least. A little like Char. Generation in an RPG.)

And pretender choice without knowing what units I get is simply an guessing game.
Even if I have to draft the pretender first in an draft-game, I still have an better guess at the units I will get, based on the possition I'm in the draft, than with the Excel.
(Look at the SadAss thread on something awful).

The Excel solution is simply too unpredictable for me.

---

Still want to see what my nation would look like, of course.

[DrPraetor]

I need to finish writing the script when I get home, but I suppose that bidding on national powers, followed by a relatively short draft, with draft orders determined by how your allocate your residual points, would work. I worry that this is going to get procedurally very complicated, but if we were going to do that, it would go thusly:

[*] First, everyone bids normalized-points (on a spreadsheet, like I'm doing now), but instead of bidding on units, spells, etc. you bid on draft positions in rounds of a draft. The things you'd still bid on (which are allocated before the draft starts) would be: Reanimation, Blood Sacrifice, Fort Schedule, Temperature Preference, Misc. National Advantages.

[*] The draft positions are determined by a proportional allocation (which you can also specify) of your residual points. So, using the same scaled-expenditure method (which I need to finish scripting up when I get home) your remaining points are distributed proportionally to the different draft orders, as you request. So a simplified system of bids might look like this:

Object/Player	A	B	C
Blood Sacrifice	200	100	60
Reanimation	0	100	50
Draft Order - Mages	800	400	445
Draft Order - Units	0	400	445

A gets blood sacrifice for 120 pts, so his bids for Mages is scaled to 880.
B gets Reanimation for 50 pts, so his bids for Mages and Units are scaled up to 475 each.
C doesn't get any powers, so his bids are scaled to 500 each.
So the draft order is - Mages: A, C, B; Units: C, B, A.
The scheme works better if there are more different rounds, of course (there's not really a "fair" way to distribute 2 powers and priority on only 2 drafts.)

[*] I really want to keep that number of drafts to a minimum, so people will be drafting a lot of stuff at once. That said, the draft orders are clearly unequal in importance, although opinion may differ about what is most important. We've got: Priest, General, Scout, Specialist, Mage x2, Cap Mage x2, Troop x4, Cap. Troop x2, Pretender, and some number of spells. That's 15+ choices, of which 6 (both Mages, both Cap. Mages, the first troop choice, and the Pretender) will I think be universally-agreed-important; most people will think the first set of mage picks more important than the second (although Frank and I don't, as it happens.) I've added "Specialists" which can get you some really sweet stuff (Mictlan Priests!), so I'll pair that with the second capital-only mage to make an unambiguously important round. Also, for people with blood sac or reanimating priests, the priest becomes an important pick, so we'll pair that with the very 1st troop (people with reanimating priests care less, because they'll be spending more money on priests/forts and less on troops, all else being equal.)

Something like this:
Round 1 - Pick First Troop (PD 20+), Pick Starting/R.E. General, Pick Recruit-everywhere Priest, Pick Nation Name, Color and Flag.

Round 2 - Pick 1st Recruit-everywhere Mage, pick First capital-only troop, pick One Spell.

Round 3 - Pick 1st Capital-only Mage, Pick Starting/R.E. Scout, pick Second Troop (PD 20+).

Round 4 - Pick 2nd Recruit-everywhere Mage, pick Third Troop(PD 1+/Starting Army), Pick One Spell.

Round 5 - Pick 2nd Capital-only Mage, Pick Misc. Commander, pick Second Capital-Only Troop.

Round 6 - Pick Pretender, Pick Fourth Troop (PD 1+/Starting Army), pick Two Spells.

Does that sound reasonable? Let me process the spreadsheets that Frank and Korwin sent me, and then we'll try a reduced spreadsheet with all the units and spells and such replaced by those draft orders, and see if normal human behavior gives fair/reasonable results?

[Korwin]

Yeah sounds more fun.

[DrPraetor]

I agree that it would be more fun - and probably more idiot resistant.

But - it can't really be automated. The nice thing about the purely spreadsheet-driven system is that I can write a script which will go all the way from loading the table of bids to spitting out a .dm file (although this is harder than I'd thought it would be.)

[Korwin]

Script not working?