D Points are a very useful measure of a skill. However that skill has quite a bit less to do with diplomacy, and much more to do with betting and predicting return on investment.

Thus a better definition of skill is needed.

From a completely subjective viewpoint, I'm fairly decent at forcing a draw and preventing solos, however pressing beyond that to securing the solo for myself is a skill I am much less proficient at. I'm also much more skilled at influencing the course of larger games (more players/more SC's) than smaller games. Thus from a pure ghostrating type measure (as over at webdip) I'm not that "skilled" a player as I don't have much in the "wins" category. Additionally the games here are so varied that a solo on Chaos would have to count for much more than a 1v1 solo. This discussion though should probably be completely separate and is likely much more related to the stats discussion in another thread for which YCHTT has very graciously volunteered to whip something up.

Thus back to solutions for CD's not worth the buy-in.
The proposed options so far:
1. Zero buy-in.
2. Reduced buy-in (say half).
Each have pros and cons. (I won't address half point buy-ins as the pros and cons are fairly straightforward there, half-measures anyone?)

Zero buy-in obviously removes the problem that a CD may not be worth the cost to pick it up, however there are potentially other significant problems. A high profile CD then becomes an incredible deal and these will be snapped up rather quickly (this isn't really a bad thing but is it fair?) Thus a player who may not be a good fit for the position (a low skill player) could pick it up, butcher the play, and still make out like a bandit even with a third of the SC's remaining.
What other unforseen and unwanted changes to game dynamics might there be for a zero point buy-in?

I've got two possible additional proposals, neither of which is a perfect solution in my mind, but maybe another mind can address what I see as the short-comings of these.
Exponentially decreasing buy-in. (Basically larger CD's cost closer to their actual value, while smaller CD's cost next-to-nothing or, when rounded down, nothing.)
This is tricky to measure though because at what point would a CD'd nation be worth its full value? Say top 25% (as far as SC count in the current standings in a game) CD's are worth full value, while the next 25% bracket is worth half value, the next worth 1/4, and the bottom 25% would be a zero-point buy in.
This addresses some of the issues with zero-point buy-in and half-point buy in but likely introduces it's own set of problems related to the arbitrary nature of "value" of taking over a CD.

My second proposal (and my preference were I to vote for one) would be a bidding system to take over CD's.
The basic idea is that you can bid anywhere from zero to full value to take over the CD. Full value would be a "buy it now" sort of option, while highest bidder (after a certain time period) would win the CD in less-than-full-value situations.
This solves all issues with trying to determine value of a CD, as each player wishing to take over a CD gets to decide how much it is worth for themselves. A 'market-driven' solution if you will.
The drawback for this method is the time period for the bidding. I know many don't wish to wait for a CD to be taken over with pauses or extends etc., but that is the only way I see to implement the bidding system. If no one bids, the game continues as normal. If someone bids less than full value, time needs to be given to allow other bids and at some point bidding will end and highest bidder wins and they take over. But when? After the phase shifts? The position will have changed, potentially drastically by that point. Thus the only way I see for this to work is to deliver what was bid on. Thus, when one or more people bid (possibly excluding "buy-it-now" for full value which could take over immediately), the bidding closes when the phase would normally have ended, and the phase gets extended by the length of the phase. Thus in a 2 day phase, if a nation goes CD, players bid to take over the CD, the winning bid takes over the CD at the end of the 2 day phase, but the phase doesn't process and 2 days are added to the phase. (As a side note if this becomes a desired solution, I think it would be a good idea to add messages to global (and also feedback in PM's to the bidding players who aren't in game yet) stating that
"player X (unnamed in anon games) has bid Y to take over nation Z" which will allow the in game players to adjust their strategy ahead of time knowing that the CD will be taken over (and at least partially addressing the "continuing with plans already in place to crush a CD" that seems all too common when a CD gets taken over). Also players in game will be able to see the phase extension coming before they get broadsided with it, especially if the phase gets extended when a player bids on the CD and the in game message could display the time that bidding ends.)

And not having read Guaroz latest point (I responded after catching up to the end of page 2, but not having read page 3) about "zero buy-in", much of what I stated above about 0 point buy-ins could be easily be replaced with 1 point buy-ins.

Oli and I will work it out and get something that gives everyone a better breakdown worked out in the next week I think. I'm just so busy at work and only have weekends and, honestly, this weekend was a busy weekend as well.

OMG, Leif, your second solution looks genius! The market will fix the "what is the right buy-in" issue! The auction could start from 1 (no gifts, please, so no zeros) and end immediately if someone bids the full worth. It would be fantastic, if it's doable. My concern about auction lenght based on phase lenght, is that it would be too short for live games (5 min?) because you could have no bids, or too long for 3, 5, 7, 10 days/phase games because the players already could wait 6, 10, 14, 20 days to have the game restarted (If I got how it works). So what about an automatic pause for the auction time? 24 hours long for all games? Most of players log in the site once a day. Nonono, it would still be not fair for lives.. so what about to get the game paused for the auction for:
1 h if phase < 1h
3 hrs if 1h < phase < 4 hrs (anyone play these? LOL)
12 hrs if 4hrs < phase < 8 hrs
24 hrs if 8hrs < phase < 48 hrs
48 hrs if phase > 48 hrs
A thing like this would allow gathering some bid for faster games, without stuck for too long the games with slower pace. Thoughts?

- - -

If Oli says auctions aren't doable, a good mix between Leif's first solution and mine could be what's following.
Consider this ratio:
X = CD worth / original bet
If iX is 3 or more I'm sure you'd say it's a high profile CD. But I may dislike to pay the whole worth, because maybe it's spring retreats (or autumn diplomacy just started) and the CD Country already has some enemy units into its SCs. Or because I fear I could be in a diplomatic disvantage (see causes 2.1 % 2.2). Or both things.
If X is less than 1, the Country in question is probably in a bad / desperate position.
So what about
TO-Buy-in = 75% * worth * X / 3

You must put that X is never higher than 3, so if it's higher then it's 3. Also that TO-Buy-in in never lower than 1 . That 75% works as a discount, to make any position appetizing because never expensive and increases the chances someone takes it over QUICKLY.

Try some example. Say original bet was 10 and try results for current worth = 1 or 3 or 8 or 10 or 12, 20, 30 (more than 30 = 30, because X can't be higher than 3).
It works!

btw, even if the auction is doable, we could use this if nobody did a bet during Auction-pause-time and the CD goes on.
___________________________________
- We're now discussing -
Players don't usually take over CDs because:
1 - CDed position are not worth the buy-in.
How do we remove this cause?

@Guaroz - I like the idea of an equation but...

The problem with any equation based on worth alone is it doesn't take into account those games in the fall/second diplomacy stage (just before a build as some games build every other years but only do one turn per year) where a nation of 9 SCs is about to lose 6 of them (one example).

And you could allow it to go all the way to 4 where the buy-in would be 100% of worth instead of capped at 75%. A classic game where a nation has 12 SCs is worth the full buy-in, assuming they aren't about to lose a great number of them.

Yes YCHTT, that's exactly why I don't like Leif's "Basically larger CD's cost closer to their actual value". Because a 20 SCs CD in a WWIV at beginning of autumn could already have 5 enemy units into its SCs you can't push back. That's the reason of the 25% discount that handles these situations and, in any case, makes the Country more appetizing (= a quicker TO)

Do you have any clue/idea on how to take into account different phases?

Well, if coded in place, the system coudl actually look at the SCs owned and see if any are already occupied byt he enemy, deducting them from the count and subsequent worth by the PPSC value.. After all, it's part of the site so has access to the full data.

Bidding is not possible. Sorry, that's to complex.

Also I'm unsure if the price is much of an concern.
Currently we have 17 open spots. 3 countries worth more than 10 .
Without the 3 highpoint games the average buy in is 3,7.
I don't think it's worth to add a complex calculation for only 18% of the open games (3 of 17), and for the other games it's not a big difference if the buy in is 4 or 3 .

Oli, as much as the bidding solution might solve the problem, the time for auction, and complexity involved in implementation are significant reasons against it. Out of curiousity how difficult is it (or how far does it deviate from standard web-dip code?)

As far as the calculation, that's why I went with a simple piecewise distribution. Here it is again with slight modification for a less complex calculation than Guaroz's equation but taking into account some phase activity:

Take the relative power of the CD'd nation (in number of SC's adjusted down by enemy occupied SC's if in autumn phase), and reduce its value based on where it falls in a proportional distribution of nations left by power (as measured by SC count. If it is in the top 25%, make it worth full value. If it is in the top half, but not the top quarter, then make it worth 50%, bottom half upper quarter would be 25% buy-in, while bottom quarter would be 1pt buy in.

Two issues both related to adjusting based on phase:
Enemy occupied SC's at time of CD aren't certain to be enemy controlled by year's end as they could be dislodged and we could create another issue.
Additionally, a nation reduced in "value" by subtracting enemy occupied SC's doesn't necessarily measure the power of the nation accurately if other nations are losing significant numbers of SC's as well.

Hence why I thought neither bidding nor piecewise categorization were "perfect" solutions but were worth discussion.

Also, we could consider a threshold. Any game for which the value is less than 5 to begin with could simply be reduced to 1 point, and such a calculation could only come into play in higher stakes games for higher profile CD's.

haha, yup, looks like I need to say 1 'p o i n t s' to have it display as 1 .

Ahhhhh! Pity bidding is not possible! :(

Leif, I'd never take over (like it happened a week or 2 ago) a 90 CD for its full worth. That's what the guys meant when they said 1.1, 1.2, 1.4, and 1.5


1 - CDed position are not worth the buy-in.
1.1 High profile CD's are too expensive
1.2 If buy in is too high, chances of reward are often slim
1.3 Many of the countries on the open list are either screwed or perceived to be in an untenable position
1.4 The idea of paying to take on someone else's mistakes is rather distasteful.
1.5 The stake is too high with no prospect.
1.6 Too probable to get a survival/defeat status. It wrecks personal stats.

I believe that even the most fantastic position you've ever seen, should not be taken over for more than 75% of his worth. We're not here to make money for the Pot, we're here to encourage TOs.

If comparison with Original Bet is too complex, I don't see how making 5 formulas (one for each phase the CD could happen) could be simpler.

I'm starting thinking that Devonian is the one, among us, whose sight was longer.

Make each TO 1 buy-in?
The quickest, the luckiest.
And....simple! :)

Sadly bidding is really too complex (I'm not a PHP-professional, I learned PHP with this site and I'm always stunned how elegant kestas programmed the main code. Most of the time I use his code and make only small (but effective) changes.
Messing too much with the code could also lead to problems once the main webdip-code gets an update.

Making the TO-price smaller is rather easy. So we should concentrate on this. (Even different calculations for each phase are ok.) But I'm unsure if this will help solving the problem. But it's worth a try.

@kaner, I doubt that question will have a simple answer.

@Guaroz, I agree that "D" Points are useful, even important, on this site. I am mostly saying that the "D" points are somewhat used as a measure of skill, and making changes to it revises the only "skill" rating system available on this site. (I am referring to "D" points as a skill rating system in a very loose sense.)

I would prefer a separate skill rating system and a "D" point system, like the Ghost Rating on WebDip. It seems like this would give the best of both worlds.

@leif, Nice suggestion on the bidding thing. Too bad it can't work.

"D Points are a very useful measure of a skill. However that skill has quite a bit less to do with diplomacy, and much more to do with betting and predicting return on investment."
- That might be true. But, I think most (but not necessarily all) Diplomacy players would want something that rated their diplomacy skills, not their skills at maximizing points on this site. Personally, I would probably want to maximize both if there were two systems. But, that's just me.

Oli is rather right. We studied the causes of the problem but we didn't study the phenomenon. I just checked it at Open Page and, among the games I was permitted to see, there's only 1 CD whose TO-price is higher than 10 , ATM. And it was 15.
This doesn't mean that 1.1, 1.2, 1.4, 1.5 are not causes anymore, because high profile CDs happen. But I think we should focus on 1.3.
Also, there were 5 open Private games: you can't TO a CD there if nobody tells you the password. But this is a different matter, not related to TO-price, and after all the game owner could be fine with it.

So I believe that the best we can do to remove causes from 1.1 to 1.5 is this:

1) Choose a threshold under which TOs are basically costless, 0 or 1 .
2) Determine a significant cut for remaining CD-prices, so that they can TOed surely and quickly, because they are "a good deal". (That's the sense of what guys said, I'm sure)

This is my proposal:
1) Threshold: any number between 14 and 24. Their TO-price is 1 .
2) Half-TO-Price for the remaining positions.
I've already explained why I don't like 0 . Yes, if a screwed position's worth 3, asking 1 (or zero) for TO is not a great encouragement, but there's not much more we can do about it, under this aspect. While if a if a screwed position's worth 10 or 12, someone could give it a try for basically no cost. About 2nd item, I believe that whatever makes unappetizing a higher profile position (Price too high even if the position looks fantastic; Concernings about the bad disvantaging diplomatic situation you may find yourself in, if you TO; Too little Dpoints to gain in relation to the high TO-price you paid to TO a winning position; You may TO a Country owning 1/3 of the board that is fighting to break a stalemate-looking line and that could be forced to a 5-way draw, losing part of the bet although you played perfectly, etc...) will disappear as you just halven the TO-price. Simple and effective. Always, I believe. Even people thinking that "The idea of paying to take on someone else's mistakes is rather distasteful" may find those positions tasteful.
Is there anything I forgot that could go wrong?

I don't understand your threshold. Between 14 and 24?

Sorry, I meant that each number you may say from 14 to 24 is fine with me, and I'd accept it without any discussion. If you propose a number outside this range, I may have objections so please provide some argument for it. I'm basically saying that the threshold would be too low if it's set under 14, and too high above 24, for the reasons you can desume from my last post. But it's just my opinion.

Maybe the buy-in is not the problem but more people are afraid of a "Defeated" in their record.

I'm confused also. Do you mean SC's or buy in cost?

Good point Oli.

Guaroz, that's not even on our list of causes, but it is definitely true with me.

@ Oli, yes, that's our point 1.6. Nobody proposed anythig about it yet. Maybe it's because many people think, as I do, that personal records are basically worthless in a site where a win in a 1v1 counts the same a win in a WWIV does. If you play a lot of 1v1, taking care of choosing an opponent weak enough each time, you'll have an incredible record. What's its worth?
However yes, many people cares of their personal record. I'm surprised nobody proposed anything yet. I got half-an-idea I have to develop. I just like to hear someone else, first. I don't know even if it would be doable...
Also, since personal record is unlinked to D-points, I'd like to close this issue before opening a new one. Of course, this does not mean we can't re-open an older closed issue if new ideas come.

@Devo: 24SCs in a Classic? :-) I'm talking about D-points, of course. About the current worth of a CDed Country. Say we decide for 18:
- if the current worth is from 1 to 18, TO-Price will be 1
- if it's 19, TO-P will be 9
- if it's 20 or 21, TO-P will be 10 ...

Uhh... Yeah, I should think before being confused next time.

Ok, I'm less confused now.

So a country that currently has a TO-P between 1-18 would in the new system (using the current example) have a TO-P of 1
19 would be 19/2 rounded down or 9, 20 or 21 halved is 10, and so on and so forth..

The question then becomes where do we put the threshold after which it switches from 1 to half current points, and why would we put it there? Do we get any other unforeseen problems with the jump from 18 becoming 1 and 19 becoming 9 ? Will people wait for a CD to drop below 18? Would we need to step up slowly from 1 point? Say 13 becomes 1, 14-15 becomes 2, 16-17 becomes 3, etc?

Also, do we make any other step functions, say make >50 a constant 25 ? Or do a continous but slower increase for greater than 50? (Thus 50-53 is 25, 54-57 is 26, 58-62 is 27, etc..

All good points, Leif, I wondered the same questions.
What I replyed to me is:
______________________________________________________
About "Half-bet with threshold" formula:

1) First of all, Countries whose worth is higher than 10 are rare, as Oli said and you can check yourself.
2) So, being most of TO-P only 1 and the rare best positions half-price, many more people than now would be checking the Open Page looking for good deals.
3) If we set the threshold at, say, 18 then yes there's a 8 "jump" at 19, but you would be paying 9 for a CD you used to pay 19 ! It looks a good deal to me, don't you think so? Would you really wait for it do drop below 19 to save further 8 ? Or would you think that in the meanwhile you're waiting, someone else could snap it up? You would probably lose your good deal. Keep in mind that more people than now would be hunting for bargains.
___________________________________________________

"do we make any other step functions, say make >50 a constant 25 ?"

Not easy to call. It may be useful for now, or for the next year long. But I believe it's not good in a site-growing perspective.
D-points in circulation now are very little because the site is young. You know there's a sort of Dpoints inflation due to the "Minimum Points" rule:
http://www.vdiplomacy.net/points.php
and due to newcomers signing up, so new Dpoints are constantly added to the circulation.
This lead to have people creating games with a bet higher than usual. Look at this ongoing gameID=2855: it's starting happen here. Average Bet-size will grow.
Until a couple of years ago, I used to play on FB Diplomacy, one of the oldest sites. Since there was no RR, people used to create 101 buy-in games, to avoid newcomers, noobs and...multiaccounters. And people joined those game normally, there were a lot of points in circulation.
They also lauched challenges like "who got cohones?" with a 1,500 buy-in:
http://speedycomputing.net/phpdiplomacy/board.php?gameID=11199
Notice the pot is not 10,500 as expected, because someone paid 699 (full price) to TO a low-profile CD. (Yes! someone bet 1,500 and then CDed!).
See... 2 years ago there were already hundreds of players owning more than 1,000 , there.
Why I told you this story?
Because, yes, it will take several years for VDip to get there, but relatively soon we could have CDs whose worth is 200 (we had a 90 a few weeks ago) and make their TO-P only 25 would mean zero.
That's why I'd say no to further "steps"
_______________________________________________

"Would we need to step up slowly from 1 point? Say 13 becomes 1, 14-15 becomes 2, 16-17 becomes 3, etc?" "Or do a continous but slower increase for greater than 50? (Thus 50-53 is 25, 54-57 is 26, 58-62 is 27, "

Well, a progressive formula would probably be a bit better. What I don't know if it's worth the effort. I wondered these questions as well, because if some day there will be games with a, say, 150 buy-in, then you would say that a CD whose worth is 40 is an almost dead screwed Country and paying 20 to TO it wouldn't be very attractive. What I replied to me is the "comparison with Original Bet" formula I proposed last Monday .

TO-P = k * worth * X / 3

Where:
X = current CD worth / original bet
If X results higher than 3, then X=3
k is a discount parameter expressed in percentage. I proposed 75% but probably 60% works better
worth is the current CD worth, so until X<3 the formula is exponential, as Leif suggested
If TO-P results lower than 1, then TO-P=1. Other results are downrounded.

Does it works? Well, let's look at an example.
Say k=60% and Original Bet = 10.
These are the results the formula gives:

Current Worth.......TakeOver-Price
_________________
from 1 to 9.................1
_________________
10 - 11 - 12 ..............2
13 - 14.......................3
_________________
15...............................4
16 - 17.......................5
18...............................6
19...............................7
20 - 21.......................8
22...............................9
23.............................10
24.............................11
25.............................12
26.............................13
27.............................14
28.............................15
29.............................16
30.............................18
_________________
31.............................18
32 - 33.....................19
34.............................20
35 - 36.....................21
...
40.............................24
50.............................30
56*............................33
*Like would be a 40 SCs Country CDed in a WWIV whose Original Bet was 10.

- In the first group there are Countries whose current worth is lower than Original Bet (10). Since they're bad/screwed positions, they're basically costless;
- In the second group regular positions: very cheap! very appetizing!
- the 3rd it's the most interesting. They're basically all looking-good positions. Whether they're really good depends on the Price you ask. Well, it grows more than proportionally from 4 for a worth of 15 (4/15=26%) to 18 for 30 (18/30=60%)
- After 30, X is higher than 3, so it's put as 3 and the TO-P is always 60% of current worth. TO is always a good deal, since they're basically winning positions.

********** If you look at the results, all questions about "continous but slower increase" or "to step up slowly from 1 point" or "Will people wait for a CD to drop below [a threshold]" just DISAPPEAR! **********


How about a different Original bet?
One of the drawbacks of the simple "Half-bet with threshold" solution is this:
You would probably consider a CDed Country whose current worth is 19 (above the threshold) like a hopeless worthless position if the Original Bet was 40. Think of a 3 SCs Country in a Modern after that all Neutral SCs were conquered. It's screwed! I wouldn't pay neither 19 nor 9!
[little thought: how rare would be this occurrence?]

Let's see what the "comparison with Original Bet" formula gives.
Say k=60% again, but now Original Bet = 40.

Current Worth.......TakeOver-Price
_________________
from 1 to 19 ............1
from 20 to 24...........2
from 25 to 28...........3
29 - 30 - 31..............4
32 - 33 - 34..............5
35 - 36 - 37..............6
38 - 39......................7
________________
40 - 41 - 42..............8
43 - 44......................9
45 - 46....................10
47 - 48....................11
49 - 50....................12
________________
51 - 52....................13
...
60............................18
70............................24
80............................32
100..........................50
120..........................72
_________________
150..........................90
175........................105
200........................120
223*......................133
*Like it would be a 40 SCs Country CDed in a WWIV whose Original Bet was 40.

- In the first group there are Countries whose current worth is lower than Original Bet (40). Screwed positions are basically costless. As the position gets better, the TO-P slowly raises!
- In the second group regular positions: very cheap! very appetizing!
- In the 3rd, they're basically all looking-good positions. Whether they're really good depends on the Price you ask. Well, it grows more than proportionally from 13 for a worth of 51 (13/51=26%) to 72 for 120 (72/120=60%)
- After 120, X is higher than 3, so it's put as 3 and the TO-P is always 60% of current worth. TO is always a good deal since they're basically winning positions.
________________________________________________________

So, keeping in mind that "Half-bet with threshold" is (now) good enough for 96% of cases, if you want to have no worries for any present or future issue and to handle even rarest occurrences, I'd say that "comparison with Original Bet" formula:

TO-P = k * worth * X / 3

would be good enough for 99.7% of cases we may ever see.

- - - - - - - -

"Half-bet with threshold" - Simple and rather flawless...now. But as the time will pass, flaws will grow as Dpoints around grow and get inflated.

"comparison with Original Bet" - Almost flawless now and forever. But not simple to explain at all.

What do we do?
Thoughts?

I wonder if a game should be canceled automatically when 2 countries nmr. Or even 1. Why shall I be forced to play a game where I cant win no matter what I do.

Uhm... So, if I'm losing a game, all I have to do is nmr and the game will be canceled. No more defeats! :)

cypeg has a good point though. NMR's ruin games. Perhaps the system would automatically send the country into CD and not process the moves?

Well, that's how it works at the moment...