Page 4 of 6

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 07:59
by Carebear
Conq wrote:(4) How did the tournament setting, 1910 cutoff, and scoring impact the game?

First, I do not like the shortened game format. This restriction is common in Europe and the Haver tournaments in America. Fortunately, most North American tournaments do not have this restriction. I do feel it impacts play, more on that later. BUT, overall I do not think it fundamentally changes the game nor who comes out in the end.

Second, scoring does impact how you think about the game. Big Gun reported the games in regards to draw size in the "Live Ticker" thread. While this is how we traditionally view games at PlayDiplomacy, it is completely irrelevant (or nearly so) at these kinds of events. Twenty or so years ago, lots of tournaments used a draw-based scoring system. However, as we know, three-ways are the most common and stable draw and it is hard to differentiate them. Most tournaments today use a scoring system that emphasizes size on the board and separation from the other players. The key is to end the game at the largest power, not tied for it, and have separation in terms of center count from the second largest power.

In looking at our ODC @ PDET first round results, it should be noted that over one-third of the games ended by 1910. So, not all games get impacted by this limitation. But, with that limitation, there is a sense of moving the game along quickly. That is, instead of playing more conservatively, there is an undeniable benefit to making an early ally, kill off the third player in your regional sphere, and then stab your ally and/or push quickly into the other sphere. The sedate and positional shifting you might see online (not in all games mind you) is lost. Stylistically, this is different, but fundamentally the game is the same and we have quickly played games online too.

At the end of the day, the scoring model still rewards solos well. In the unrestricted events, like last year's WDC, we still see a reasonable number of solos. However, I think the biggest difference is ensuring you are board leader at the end. In an online game, you might eschew being board leader for most of the game to keep the target off your back while you position for the stab and the run to a solo. In these tournament games, you have to walk more of a tight rope because at the end of the day you need to be board leader while still avoiding being targeted.

So, while I do feel these impact game stylistically, I do not think they fundamentally change what you need to do to be successful.

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 08:36
by Carebear
Conq wrote:(5) If you are an unknown newcomer, how did that impact your games? Was it an advantage or a disadvantage to be an unknown?

See below.

prgm wrote:I think being a newcomer was a disadvantage but that may only have been my experience. I am sure that the other attendees from this site will give their own views in due course when they get back home. Some had a far greater distance to travel than me.

asudevil wrote:Always my fear and with 2 people expressing similar...turns me off to major desire to planning for DC

Machiara wrote:This almost completely removes any desire I might have to make the effort to attend one of these events myself.

I am going to shut this down hard. I apologize ahead of time for being so forceful.

First and foremost, skilled players without FtF traveling experience can be very successful at these events. In particular, Tanya Gill placed third at the WDC. She is a webDiplomacy regular with the name DemonOverlord and goes by Durga here. IIRC, she has maybe played a "house game" or so before, but has never been to any tournament. Now, I might express an opinion that her being female and attractive might have helped her a bit -- it certainly would not have helped at all had she been less than a strong player. Likewise, Mikalis Kamaritis had two very respectable results with the same background (another webDip regular that goes by Hellenic Riot there and here).

Further, while a new player may be an unknown, as long as you show competence you will get a chance. The players that do know each other oftentimes have memories of painful stabs. Besides, known players are also a threat for board top. It is quite common for a known player to prefer a competent newcomer to ally with early to rid themselves of a known strong adversary and then stab the newcomer later.

Unfortunately and to be perfectly frank, I think prgm's experience is more likely about him needing to improve his game. I enjoyed meeting him; he was very nice and personable guy. But, if you look at his PureDip results, he is only drawing in about a third of his games. Unfortunately, did not get a chance to play him to maybe be able to provide some feedback.

To conclude this part, if you are really good, you will get results regardless of whether it is your first tournament or not.

prgm wrote:It was a positive experience for me, was good to try and therefore I would recommend it to others. I especially enjoyed meeting the other members of the Play Diplomacy site - we had a good laugh and they were good company.

@asudevil and Machiara, this is the real point. Regardless of the success on the board, prgm still felt it was a positive experience and recommends it to others. Meeting other PlayDippers AND other members of the hobby is a great experience.

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 10:20
by Carebear
Now for the AARs... Mostly this is from memory as I did not take games notes afterwards as I sometimes will.

I played in the two rounds on Friday, skipped the morning round on Saturday, played the evening round on Saturday, and played in the final round on Sunday that was part of the team event.

Round 1 Board 4
Austria: Buachaille
England: Peter Lund (peterlund)
France: Doc Binder
Germany: Mikalis Kamaritis (Hellenic Riot)
Italy: Marcus Edholm (Ezio)
Russia: Jed Cotgrove
Turkey: Carebear

Opening Thoughts
I hate Turkey more than any other power. It is so easy to keep it in check. :( Doc Binder is a past WDC champion. I have played with him before on several boards in the past. He was the strongest player on the board. Normally, I like to be adjacent to the best player as I can generally work with and grow with them. Unfortunately, with him on the opposite side I would have to work hard to keep pace with him. The first impression of the three webDiplomacy players was that they were not going to be especially troublesome. Russia likewise. I felt having Buachaille next to me was a mixed blessing. I knew he would be a decent player, but was unsure about how he would handle this environment.

Summary of Play
Sided with Austria early and made quick work of Russia. Wanted to get Italy on board with a triple, but he would not move against France (Doc). With the south not keeping France in check, he worked with Germany to kill off England and quickly became board leader. Tried repeatedly to get Austria to move against Italy quickly since Italy was basically doing nothing, but Austria was very conservative. Worked with Russia as my vassal to help stave off German incursion into the east. Finally, got Austria to move against Italy. But, by that time France was huge and already in the Med. :( As the final stalemate position looked to be probable, I stabbed Austria to move from fourth to second on the center count. Austria went from eight centers to three. Though, he really messed up several moves that would have gotten him two more centers, notably Munich that might have given us(me) more life to get closer to France center count wise.

In the end, we had Ionian stalemated and the traditional line along Germany and Austria. Austria could have thrown to France/Germany, but that would have reduced his standing to almost no points. So, it was worth it to him to still work with me to hold the line. Germany wanted second place and squashed the draw vote. I was able to use this to leverage France to stab him to end the game. France stabbed and Germany accepted the draw. France had twelve to my ten centers.

Post Game Analysis
Frankly, given the board dynamics this was what I expected, second to Doc. Austria played too conservatively for us to keep pace with the west. :( He needed to take my advice to attack Italy earlier and move on France. If that had happened, I might not have felt the need to stab him later for board positioning in the endgame. I got a solid score from this game.

Round 2 Board 3
Austria: Carebear
England: Demis Hassabis
France: Dave Maletsky
Germany: Rene Van Rooijen
Italy: Emmanuel du Pontavice
Russia: Ben Barber
Turkey: Philipp Weissert

Opening Thoughts
Ugh, Austria. Last year's WDC top table player as France. :( Phil as Turkey (finished sixth this year). :( Emmanuel as Italy. :( Can I survive???

Summary of Play
Turkey really wanted to be my friend. Unless I am Turkey, I hate to see Turkey grow. But, as he was the most enthusiastic I worked with him to start. Would have loved to seen Italy go west, but he was having none of it. Then, I might have move against Turkey earlier. Instead took Venice and pinned Italy back. Helped Turkey into Rumania. Again saw France playing England and Germany against each other and having a grand time in the west. :( Decided to try and make something happen sooner and tried to stab Turkey. Totally flubbed it, as I did not communicate with Russia. However, Russia at that point was just happy to have a friend and in effect became my vassal for the rest of the game. I basically stagnated holding Italy and Turkey while France grew and came south. Offered to help France in exchange for friendship, had no choice really. Became his vassal. Came in second alone again. But, had an awful score as he ran away with 14-16 centers to my six. :(

Post Game Analysis
This was a tough board with tough players around me. But, I still misplayed it. I should have stuck with Turkey longer. Phil came up to me afterwards and asked what he could have done. All I could tell him was that I don't like Turkey. Frankly, that should not have affected me as much as it did. The flubbed stab all but ended it for me too. :( Worst score of the event.

Round 4 Board 7
Austria: Dirk Brueggeman
England: Alex Lebedev
France: Carebear
Germany: Marvin Fried (Jamesturner9000)
Italy: Ivan Woodward
Russia: David Percik
Turkey: Ruben Sanchez

Opening Thoughts
Hey, finally a decent power!!! Ugh, the tooughest board of my tournament. Germany was current tournament leader, eventual top board player, and finished second this year. England was also very tough, he finished ninth at WDC this year. Italy was another solid experienced player. Turkey was another top player and played on the WDC top board this year again.

Summary of Play
Germany and I hit it off and Italy wanted to head east. Great, despite the extremely tough board maybe I can get something. Russia and England got into it early. Made it simple. Germany and I built double fleet in W'01. Made a sweet tactical move in F'02 that got kudos from Germany and messed up the English defense. Italy and Turkey unfortunately were also making quick work of Austria. As Germany, Russia, and I were finishing up England, I noticed that Italy was in position to stab me at least. I requested Germany to allow Pic - Bur even though he had nothing in Mun. He realized he was vulnerable too, but had not anticipated the Italian turn. I saved our bacon. Italy stabbed and made progress in both Germany and France. But, I helped Germany to take Munich and hold the line. England's last fuck you was to break my line against Italy in the MAO. Fortunately, Turkey took that moment to eviscerate Italy in Austria. Having kept good lines of communication open with Italy, we negotiated a staged withdrawal so that he could get his forces back to slow Turkey. I let him keep Marseilles for a year. I had played very well tactically up to then. It was at that moment that fatigue caused me to not put the coast on the fleet move to Spain. Not terribly damaging, but a small tempo cost.

Identifying that Ruben would now want a draw. I discussed endings with Germany. He said he would openly veto a draw vote, take the heat, and did not mind if I grew to second alone (German, Russia, and myself were all tied at six) or even challenge for first. They were along the stalemate lines and had little growth potential while I had a completely open Marseilles that was ceded to me by Italy plus future Italian centers as desired. Great! Remember, this is how Germany mis-negotiated with Doc in my first round game.

Ruben called for a draw, Germany forgot(?) to openly veto, and we voted. No problem, I can blackball it. I will be assured second alone and can close the gap with Ruben for another decent score. The TD and regular assistants were busy. We got a secondary guy. He named the colors for the stop and continue playing. They were opposite normal, but fine. I selected. Before voting, someone spoke up about switching the colors and he said yeah that makes sense and switched them. I said okay -- but in my fatigue did not swap colors. :( I dropped a stop instead of continue. Game ended. I was in shock. Italy and Germany were too as they expected me to blackball it as I had growth to play with. :( What a cock-up.

Post Game Analysis
I actually played quite well this game. I got positive feedback from a number of players about my tactical decisions. And, this was a very tough board. Unfortunately, fatigue played a key role in my ruining my score for this one.

Will do my last game in another post in another day or so -- as a prelim I was board topping but got stabbed and it was my fault. :(

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 16:43
by StarWatcher009
Hey Carebear,

Was your board 2 England the same Demis Hassabis that runs DeepMind?

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 17:19
by Carebear
StarWatcher009 wrote:Hey Carebear,

Was your board 2 England the same Demis Hassabis that runs DeepMind?

Doh! Yes! There were multiple people from DeepMind there.

Talked with another person from DeepMind Thursday evening about making Diplomacy AI the next target. He talked about introducing me to the CEO. Didn't happen. But didn't know at the time that was him!

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 18:24
by StarWatcher009
Carebear wrote:
StarWatcher009 wrote:Hey Carebear,

Was your board 2 England the same Demis Hassabis that runs DeepMind?

Doh! Yes! There were multiple people from DeepMind there.

Talked with another person from DeepMind Thursday evening about making Diplomacy AI the next target. He talked about introducing me to the CEO. Didn't happen. But didn't know at the time that was him!

I'd wonder what his approach to a diplomacy AI would be. It would have to be some combination of text recognition/response as well as the tactical side which is probably pretty similar to Go.

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 20:11
by DQ
StarWatcher009 wrote:
I'd wonder what his approach to a diplomacy AI would be. It would have to be some combination of text recognition/response as well as the tactical side which is probably pretty similar to Go.

I spoke with him very briefly on the subject, he argues it is a much, much more difficult problem than any approached by AI to date (for reasons we all understand), but that as such, its a really juicy target.

Re: WDC 2017: AAR and General Impressions

PostPosted: 17 Jul 2017, 23:11
by StarWatcher009
DQ wrote:
StarWatcher009 wrote:
I'd wonder what his approach to a diplomacy AI would be. It would have to be some combination of text recognition/response as well as the tactical side which is probably pretty similar to Go.

I spoke with him very briefly on the subject, he argues it is a much, much more difficult problem than any approached by AI to date (for reasons we all understand), but that as such, its a really juicy target.

Yes. I think DeepMind are way ahead of where we think they are though. I've spoken to him before and he suggested as much. Thus I think they could cobble together a fairly good AI pretty quickly, especially if many of them are going to WDC.

Who knows, maybe next years PDET will have an AI participating.

Re: WDC 2017: AAR and General Impressions

PostPosted: 18 Jul 2017, 06:21
by Big Gun
This comes in a little later than I had hoped, mainly because very shortly after WDC I flew out to Guadalajara/Mexico, where I’m currently running a summer camp for kids.

So busy times right now, but WDC is still etched in my mind as if it was only yesterday. Like WarSmith, I’ve lost quite a bit of sleep by reliving my games over and over and there are still times – 8 days later – where I stop to kick myself that I didn’t go for the juggernaut in the name of non-metagaming or jump at a 3-dot stab on England in the team round! O, well.

The advantage of commenting late in this thread is most of the things I wanted to say have been covered by others by now.

So just a few points here and a short summary of the four games I played.

My one and only experience playing in a F2F tournament was in the mid 80s at MidCon in Birmingham, UK. I was just 19 or 20 years old. In those days I was a typical unstable, impetuous and easily influenced young whippersnapper and believed I knew everything. I didn’t do well. Subsequently I stopped playing Diplomacy for 30 years and returned to the game only about two years ago, but playing only online on this site.

So going to WDC felt like the right thing to do. I was very curious about the whole adventure and was particularly excited to meet up with players in real life from this site, some of whom I had played with online.

The whole tournament was enormously rewarding. Playing F2F is exhilarating, intense and thrilling in a way that online can never come close to emulating, even with our online “live” games on 15-minute turns.

A lot has been said about the time pressure of F2F and that is probably the most decisive difference. The 15-minute time frame per season is nearly always shorter, as it includes the time required to adjudicate the previous season’s orders and make retreats and/or builds. So you might realistically only have 11 - 12 minutes in which to write down your units, consider the board, try to grab the 2 or 3 or more players you need to talk with and have meaningful conversations, agree on a plan, then write your orders and pop them in the box. This whole process is initially sheer stress for online players, who normally enjoy at least 12 hours to deal with these issues and often a lot more. But it’s a process you get used to over a tournament of 4 F2F games and I felt more relaxed as the tournament progressed, although my adrenalin-filled excitement was still fueled by the general atmosphere, with between 9 and 11 boards running simultaneously.

My first game didn’t go well. I played Russia. England told me in Spring 01 that he intended to play north and made it clear that he wasn’t prepared to negotiate. So, allied with Austria, I opened north and bounced England out of Norway in Fall 01. By 02, I was on 7 units, but with an army in Norway and Rumania and a useless single fleet in Sevastopol, without a hope of getting into the Black Sea. I’d positioned myself awkwardly and a very strong Germany took full advantage, using England as his puppet to cut me in half. The game showed me how easy it is to make positional errors early on, from which you can hardly recover at all. I was eliminated in 07, deservedly so.

In my second game I drew Italy. I spent a long time on 4 units, a very typical syndrome for Italy, I think. Turkey was hostile throughout the game and lied incessantly at every turn. Although he was at one stage in the Ionian, I managed to dislodge him, narrowly avoiding what would have been a fatal convoy to Apulia. I ended the game with what I felt was a respectable 7 units, 2nd I believe to Germany.

My third game was perhaps the most interesting of all. I drew Austria and found myself playing side by side with former World Champion Chris Martin, who drew Italy. We quickly found a good rapport and began an alliance that was to hold for all 10 game years. Turkey defended well, so we started a kind of delayed Key Lepanto that enabled Chris to build a 5th and then 6th unit, while I was at 5. At this stage we realized we were facing a very strong Western Triple, which had emerged around 03/04 so for many years we had to play defensively, whilst keeping Turkey in check and watching Russia’s slow demise. In 1908 England finally stabbed Germany and opened up the can of worms we’d been waiting for. But it was too late to make any significant growth and in the spring of 1910 Chris was on 7 and I had 6, with England clearly leading on 10, but losing ground. At this stage I realized Chris might still have a shot at a place on the Top Table, if he could get from 7 up to 11 by the fall and finish as board leader. So – after playing a respectful 9 years together and with nothing much to lose from my side – I offered Chris the only 2 of centers of mine, which he could take and he also had a shot at 2 of Turkey’s centers too. Yeah, I know, I can already hear those loud screams of “Kingmaker” and it’s true, but I am not ashamed to admit it. My alliance with Chris was the only long-term alliance I had in the whole tournament and it felt like an honorable gesture. Unfortunately it didn’t quite work out for Chris as Turkey once again guessed well, so Chris finished with 9, not quite enough to reach the Top Table. I ended with just 4, but felt happy with the result nevertheless.

My fourth game as Germany didn’t go well at all. Once again I positioned myself awkwardly early on. My alliance with England was strong but France defended against us easily and for quite a while, which simply meant Austria (Chris Martin again) and Russia could gain enough ground to stab me together. I was eliminated in Fall 1910 which is kind of bitter, but again, I made some horrendous mistakes. I’m happy to say that I feel I’ve learned something from them. I also spent about 4 game years with just 1 or 2 units, which enabled me to watch and listen to some fine diplomacy amongst the other players on the board.

Playing with the elite of the F2F crowd was a real eye-opener. These players are experts at what they do. It was fascinating to watch some of the top players’ body language, facial expressions, gestures, eyes – and listen to the way they phrased things cleverly, sandwiching their sales talk with heavily buttered toast and not batting an eyelid at stabbing or being stabbed.

Also – I realized that you have to be very much an expert with the mechanics of the game and understand tactics quickly. Top players can read any board in 2 minutes and know what to do. That’s quite an art.

Despite my less than satisfactory results, I loved every minute of WDC 2017 and can only recommend it. It’s given me a serious kick in the backside to study the standard game again in greater depth, after spending so much time on this site playing on the Versailles map.

But – like nearly everyone here on this thread – I would like to underline that it is the people that made this tournament great. No one was unapproachable, the social element and community feeling was very strong and newbies did not have a hard time making friends. I made loads.

Those were 4 really happy days in my life. Besides the games, the absolute highlight for me was meeting the PlayDip players. I particularly enjoyed scoffing curry and skulling pints with WarSmith and GPD, sharing banter and whisky with Buachallie and listening to Carebear’s blow-by-blow accounts of a game over a greasy full-English breakfast.

I am definitely going to play more F2F in future.

Cheers. Big Gun

Re: WDC 2017: AAR and General Impressions

PostPosted: 18 Jul 2017, 11:24
by Captainmeme
Just some thoughts on the 1910 cut off...

I actually think that that's a good thing for a FTF game, although there are better ways of doing it. At NDC 2015, the games had no cutoff year and it was honestly really tiring. Playing a FTF game to a stalemate or solo is often not fun for anyone.

However, specific cutoff years are also not good because they tend to make the last couple of years SC rushes, with no thought for consequences past the cutoff year (since those are not being played). In my mind, a good system tries to encourage players to play as if the game would go on, without actually making them play it out.

By far the best system for that is the unspecified cutoff. The TD secretly chooses a year between 1909 and 1912 for that rounds games to end, but the players do not know what year it will be until the TD announces that the games have ended. Players will still go for late SCs if they can, but they'll be much more wary of the consequences.

I also need to talk about scoring systems. Some people dislike draw scores being centre based in FTF rather than draw size based. I'd argue that having centre based systems for FTF is an absolute necessity, and I'm also of the (generally unpopular, admittedly) opinion that these systems should be standard for online games. Let me explain...

As mentioned above, I prefer that FTF games are cutoff versions of what the game would be if it was played out. It needs to end early so people aren't playing the same game all day, but the scoring system should be designed to encourage people to play as if they're setting up for a solo victory, even if time does not allow them to actually reach that solo.

At this point, we need to define what 'playing for a solo means. There are a lot of ways to solo a Diplomacy game, and it's incredibly difficult to get a scoring system to represent that, so we use the most common factors that allow high level players to solo high level games:

1) Have a high centre count. This one seems obvious - it's much easier to solo with 14 units on the board than it is with 9. Since we have no way of looking at positioning when scoring games, this is the best way to approximate whether they have the ability to solo through their tactics on-board.

2) Keep the opposition as divided as possible. This actually aids the soloing power threefold - first, there will not be another huge power, so there's no need to call off a solo attempt to stop someone else's; second, if the opposition unite to attempt to prevent the solo, they will be much less coordinated if 6 players have to work together than if just 2 have to or if 1 player has the majority of the units; and finally, most soloes occur because someone else wants them to. By keeping more players in the game, the soloing power massively increases their chances of someone in that enemy group wanting to throw the game.

The best way we account for this is a Scoring System called 'Sum of Squares'. I won't go into the formula (although it's pretty simple) but the basics are that you will always benefit from taking SCs, but you'll lose points for other players growing above the average SC count. The objective in sum of squares is to take as many SCs as possible, while keeping the other players roughly even on centre count and not letting people catch you up. It matches up pretty well to the two solo conditions listed above.

Of course, there are problems. Some of them relate to positioning - our scoring system cannot detect positioning, so a clearly stalemated 16 centre Turkey will be overvalued, and an Italy with just 9 centers but poised to take the lowlands and the Russian homes will be undervalued. This is unfixable - there's no way to make a scoring system take this into account unless you introduce human judges to independently score each game, or we manage to get an AI capable of playing the game out, neither of which are feasible right now.

There are fixable problems too, though. A quirk of the Sum of Squares formula is that there is practically no difference between a 1 or 2 centre draw and an elimination. This is rightly criticised as against the spirit of the game, as you should always have a reasonable incentive to try to work your way into a draw and keep your nation alive, regardless of how small you might be in said draw. This can be fixed by giving a 'Survival Bonus' of ten points or so for making it into the draw, thus making it much better to stay alive than to be eliminated.

The other big issue is that this scoring system disincentivises the 'Alliance Play' method of soloing - namely, convincing a player that you'll two way draw with them and rolling across the board with them, only to stab them last minute to get to 18. This is a legitimate way of soloing, although it's seen much less in high level games because good players tend to see through it. At the end of the day, a system that incentivises both this and the above factors could not be found, so it was decided that it's better to incentivises the method of soloing that works most often in high level games, as opposed to Alliance Play, which is much better against newer players, on the basis that if you're scoring a game in person it will generally be a high level game.

Still, a small incentive is still given for Alliance Play - gaining a lot of SCs is still rewarded, albeit less if someone else is also gaining, but also you can add a 'board top' point reward. The reasoning behind this is that, if you were going to convince your ally to let you solo, you should also be able to convince them to let you top the board.

Combine these improvements to Sum of Squares and you essentially get Manorcon scoring, which I will forever argue is the best scoring system. It rewards you most for achieving a solo, and if not then how close you got to a solo, and finally rewards you for making it into the draw over a defeat or being soloed on. That's exactly what a Diplomacy scoring system should do - especially one for games that end early.

Part II - Sum of Squares in online play - coming soon :)