This is a belated summary thread of a recent tournament I hosted for the purposes of beta-testing a proof-of-concept template featuring the luck setting at 100%, to see if such a template can actually be fun and strategic at the same time. The tournament was called "Pure Luck" Strategic 1v1 (multi-delay) Beta-test: https://www.warlight.net/MultiPlayer/Tournament?ID=16560

First of all, I would like to thank all the participants for helping to beta-test this template. I very much appreciate it!

Next, I will discuss the background and origin of the template idea, the specific settings that were used, some controversy surrounding these settings, how the tournament was set up, and finally the results of the tournament. (In a future post I will analyze the tournament games in more detail, hopefully with some useful numerical analyses.)

Background

The idea for this "Pure Luck" Strategic template came when szeweningen ran his 1v1 Template Contest (https://www.warlight.net/Forum/117170-1v1-template-contest). I noticed that most of the early proposals for templates involved very low luck settings, and at least one person explicitly suggested to restrict the possible settings to no more than a certain luck percentage (such as 16%). I thought this was strange because I had previously seen discussions in the forum where there was no clear consensus that low-luck-rate necessarily equates to more-strategic-play. So, playing Devil's Advocate, I asked, "Do you consider luck % as being necessarily a negative? If I made a template with 100% luck involved would it have a lower chance of winning?"

After posting that, I re-read some of the thread, and it appeared to me that the host, szeweningen, had already declared that he would not impose such restrictions on settings, and that using unusual settings would not necessarily influence the scores, when he said:

I really don't want to put in specific rules for settings, i'll let the creators decide what they think is strategic and what is not. The reason behind it is that there are some well-established strategic settings that generally will work on every map, but possibly some things we think of that are not strategic on a general principle, may work in a very specific situation. Different kill rates, card focused 1v1 template, low base income, limited distribution, fog settings, these are all relatively unexplored. Maybe there will be some other new and original ideas or a mixture of those. I'm open to be surprised, if I can I'll test all the templates myself.

However, several people responded in the thread arguing that 100% luck could not be considered strategic. It was this reaction that gave me the idea to come up with a proof-of-concept template that would show that 100% luck (or 'pure luck' as compared to 0% luck being called 'pure skill') could indeed be played in a strategic way, according to the guidelines szeweningen had proposed, that "possibly some things we think of that are not strategic on a general principle, may work in a very specific situation".

The key principle that allows 100% luck to remain strategic is the well-known probability theory result called the Law of Large Numbers (https://en.wikipedia.org/wiki/Law_of_large_numbers). The more times you flip a coin, the more likely that the total proportion of heads approaches 50%. With 4 coin flips, there's still a 1 in 8 chance that you'll get either 0% heads or 100% heads, but with 4 million coin flips, the chances that you get significantly more or less than 50% heads (for a fair, unbiased coin), is very very small.

Well, in Warlight, each attacking or defending army is like a coin who gets flipped, but instead of being an even 50% chance, the game uses the default 'kill rates' of 60% for attackers and 70% for defenders. With only a few coins/armies, it's not unusual to get 100% heads/kills or 0% heads/kills. But with many more armies, the chances of getting such extreme 'unlucky' results decreases rapidly.

With this in mind, the design of the "Pure Luck" strategic template was relatively easy.

The Template Settings

Since the idea for the initial template was only to be a proof-of-concept, I did not put any real thought into trying to actually win the template contest. Therefore, I opted to make as simple a template as possible that could demonstrate how 100% luck could still be played strategically. I started with the widely played 1v1 strategic template (used in auto-games and the 1v1 ladder), boosted the luck setting to 100%, and then changed all values for armies to 10 times their original amount: starting armies at 40, neutral armies at 20, wasteland armies at 100, base income at 50, every bonus value multiplied by 10, and the reinforcement card value at 50.

I wondered to myself why Warlight didn't already use this idea, and it didn't take long to think of a likely reason: With so many armies on the map, what's to stop someone from using a whole bunch of them as single-army delay moves to try to out-delay the opponent each turn, which in a competitive game can provide a huge advantage. But if everybody tries to out-delay each other, then the game bogs down into dozens if not hundreds of single-army delay moves. And who wants to play something so tedious like that?

However, it also didn't take very long to come up with a possible solution, which involves the use of multiple Order Delay cards, which will be described below under the 'Tournament Setup' section.

I submitted my initial template to the contest with this introduction:

It's an attempt at a proof-of-concept that 100% luck WR could theoretically still be usable in a strategic way. I upped all the armies by 10 times, so that should reduce the standard deviation quite a bit. You may get a few odd results, but it's much less likely than if you were playing with the standard income and distribution settings. (10 armies might occasionally fail to take a 1, but 100 armies will almost surely take a 10.)

I also posted some sample games and mentioned my thoughts about the potential issue with delay moves, and how this could probably be solved with Order Delay cards:

Gameplay is very similar to the standard 1v1, just with a bit more variety. Also, you can pretty much delay as much as you want. I'm thinking of adding in more delay card pieces to reduce the impulse to use delay moves. Instead, you'll have to decide how many delay cards to use or to save for later.

Controversy

Several contest participants, including myself, were disappointed when the hosts of the contest live stream did not seem at all interested in such concept templates, despite the fact that szeweningen's early posts seemed to indicate that concept templates were what was desired. I believe that was simply the result of unfortunate miscommunication and misunderstanding of szeweningen's intent for the contest. (This initial controversy begins on the thread at https://www.warlight.net/Forum/117170-1v1-template-contest?Offset=80 at 11/30/2015 15:24:54.)

For the 'pure luck' template, however, there was still clearly a misunderstanding of how 100% luck could be strategic, and even a claim by szeweningen and others that the potential delay move problem was a fundamental and fatal flaw in the concept of the template, which could not be resolved by using Order Delay cards.

However, I was very surprised that no one had even tried the template, and neither had they even bothered to look at any of the 7 example games I had provided, where delay moves were clearly not an issue.

A long discussion ensued on that thread (starting on https://www.warlight.net/Forum/117170-1v1-template-contest?Offset=100 at 12/1/2015 03:39:06), which resulted, from my point of view, in the people opposed to 'pure luck' strategic simply declaring that delay moves would ruin the games and that using Order Delay cards, as I had proposed, could not possibly fix the problem. Still none of them had tried the template themselves to see if their intuitions were right or wrong.

It was at this point that my suggestion that such a dispute of opinion could be solved by actually testing the template in a tournament led me to actually start such a tournament just to test it myself, mainly for my own curiosity, but also with the idea in mind that I would publicly post the results of the tournament, regardless of whatever results occurred.

This thread (this initial post plus its continuation below), therefore, is the first part of that resolution, where I only summarize the tournament and its basic results. The second part, forthcoming, is where I will attempt to analyze the actual games of the tournament and produce some numerical results, such as: How many games were heavily influenced by just good/bad luck? and How many small delay moves actually made any difference in the results of a turn?

[cont'd...]

Edited 2/28/2016 06:19:39

Tournament Setup

The main change in the template used for the tournament was the introduction of 'multi-delay' cards, which essentially just means that Order Delay cards are very plentiful in this game, rather than being relatively rare as in most games. In this new version of the template, each player recieves 1 whole Order Delay card per turn, and they even start with an Order Delay card from the beginning (to account for the not unusual case where you start next to your opponent, or want to attack a neutral that your opponent might also be able to attack). The reasoning is that if each player has the opportunity to play 0, 1, 2, 3 or more delay cards in a turn, then using small delay moves becomes much less useful, because it's quite likely that your opponent might either a) simply play one or more Order Delay cards more than you do, or b) choose not to play any OD cards this turn, but save some up for a later turn, where it might be more important.

In fact, the only time that small delay moves will make a difference is if you and your opponent just happen to play exactly the same number of OD cards in the very same turn. If either player plays more OD cards than the other, then in that circumstance any small delay moves by their opponent will have no practical effect on the order of major attacks at the end of the turn. The more common OD cards are, the less likely it is that the players will play exactly the same number of them in any given turn. Of course, you don't want to have hundreds of OD cards, because then it just becomes a tedious game of playing hundreds of OD cards, rather than hundreds of small delay moves. And who wants to play such a tedious game? I figured 1 card per turn seemed a reasonable compromise, and so that's what I went with. It's possible that more or fewer might actually make for a better game; I don't know.

The tournament was a round-robin tournament, which I quickly learned is actually quite an unpopular format for mid- to large-sized tourneys. I was considering switching it to a double- or single-elimination tournament because of the feedback, but fortunately enough players joined to fill up the 10 slots, so it actually started fine, without a hitch.

Since one of the objections I had received was that only newbies would not find themselves forced to play dozens of delay moves, because they are not as competitive as the top players, I decided I should primarily invite players who are in the top ranks of the 1v1 ladder.

(Here again is the tournament link to see the settings and players: https://www.warlight.net/MultiPlayer/Tournament?ID=16560)

The players, in clockwise order starting from the 3:00 position were:

• PhucilliJerry
• John Smith
• HyperionKnight
• Robert E Lee
• wct (myself)
• ntw0n
• fireice82
• Ender2010
• Sułtan Kosmitów
• Ranarius

7 out of 10 of these players, including myself, were at the time ranked higher than 1700, which is currently the rating of roughly the top 100 players on the ladder. (Believe it or not, at least one person has actually argued that most of these players are still newbies, despite being in the top 100 of the 1v1 ladder, which equates to approximately the top 20-25% of ranked, active players on the ladder.)

I myself made a personal commitment to not play any small (1-3 armies) delay moves, and rely solely on Order Delay cards, and 'normal' leftovers-style delay moves for delays.

Results

The final rankings of the players are as follows:

1. (6W 3L) Robert E Lee
2. (6W 3L) PhucilliJerry
3. (6W 3L) fireice82
4. (5W 4L) wct
5. (5W 4L) Ender 2010
6. (5W 4L) ntw0n
7. (5W 4L) HyperionKnight
8. (4W 5L) Sułtan Kosmitów
9. (2W 7L) John Smith
10. (1W 8L) Ranarius

Congratulations to Robert E Lee! Honourable mentions go to PhucilliJerry and fireice82 who matched him in wins/losses, but were ranked lower because Robert E Lee defeated both of them in their individual match-ups in the round-robin format.

One unexpected result was that most players got between 4 to 6 wins, and the highest rated 1v1 ladder player, Sułtan Kosmitów, only won 4 of his games, ranking only 8th. This may be the result of the luck setting being simply too high, but an alternative plausible explanation, which is my actual opinion, is that these unexpected 'flat' rankings are due to the fact that most people in the tournament were simply unused to playing with these particular settings (large numbers of armies, many delay cards, 100% luck). As a result, especially in the early games, some weak play from strong players cost them a loss or two instead of a win or two. My hypothesis is that if these same players played this template again, they would fare much better, and much closer to their 1v1 ladder rankings. However, this current post will not attempt to answer this question; the follow-up post with more detailed analysis will do that.

What is most intriguing, however, is that if you actually open some of those games and see how they play out, small delay moves seem to make virtually zero impact on the results of the games. Again, this post will not address that definitively; the future post will; but these games are available for anyone to examine and scrutinize, and I doubt you'll find more than one or two instances (if any) of small (1-3 armies) delay moves making any difference in the game.

For me, this already is enough to convince me that a) 100% luck can indeed be strategic, and b) small delay moves are not actually a problem; plentiful Order Delay cards virtually eliminate their value, even for competitive players. However, I think it is still up in the air whether c) the template is actually fun to play.

Personally, I found the tournament stressful to play. But I'm unsure how much of that stress was due to a) the increased thinking complexity required to decide "Do I deploy 35, 36, 37, ..., 43, or 44 here?" vs. "Do I deploy 3 or 4 here?", b) the very high skill of my competition (I have also recently found the 1v1 and 2v2 ladders too stressful to continue playing, and have quit them for the time being, largely due to RL stuff happening that I have to deal with), c) my own internal hopes/biases/expectations about the outcome of the tournament, d) simply the stress of the extra RL stuff I have to deal with these days.

I haven't heard much back from the other players yet on whether they found the template fun or not, although I did get a couple comments in a couple games saying they prefer the standard settings, which isn't quite the same thing as saying these settings were 'not fun', but may hint in that direction.

One thing I didn't hear any complaints about, however, was anything about small delay moves ruining the game.

I encourage anyone interested to try out the template (or a variation of your own) to see how it plays for you. Seeing for oneself is much better than taking someone else's word for it, in my opinion.

Here's a link to the tournament's template: https://www.warlight.net/MultiPlayer?TemplateID=759977

Edited 2/28/2016 05:51:51

It's a great template, a lot of the concerns turned out to be non-issues, and on the whole I found the template very enjoyable to play :)

Thanks for the feedback, Fleece. :-)

I didn't play the template, but the concept seems totally sound to me.

(By the way, Warlight Statistics keep track of "luck" during games, so it's possible to very directly look at how "luck" factors in to wins and losses.)

I just wanted to say what a pleasure it is to read such a well-written and well thought-out thread. I wish there were more like this on the forums! Thanks.

Interesting concept.
This seems to show that expansion diplo templates could potentially be strategic(hmm, that's a weird thought).

(By the way, Warlight Statistics keep track of "luck" during games, so it's possible to very directly look at how "luck" factors in to wins and losses.)

I hope to use those very stats to help my analyses, especially in relation to 'how many games hinged on good/bad luck?' :-)

I just wanted to say what a pleasure it is to read such a well-written and well thought-out thread. I wish there were more like this on the forums! Thanks.

Wow, thanks. :-) I was just trying to make sure I covered all the points and objections that were brought up. It did take some time to put it all together and organize it. Glad to hear it's readable.

I only barely have time to skim a little, but this is a very interesting (pair of) posts :)))

right now im just rooting for everything multiplied by 100 games lol

Right. I don't see a reason why you wouldn't be able to play just as easily with x100 or x1000. Well, maybe the numbers would start taking up too much room on the maps. Also, maybe WL's random number generators might have to do a lot more work (although there are techniques Fizzer could use to streamline the binomial distribution generation, if he's not using them already).

I should mention, however, that there is one other possible undesirable side-effect of having so many OD cards around that I just thought of today: If basically every important turn ends with one person playing more OD cards than the other, then you can get into a situation where "either all my important moves happen before all of my opponent's, or all my opponent's important moves happen before all of mine," rather than the usual way WL handles it where both players' moves are inter-leaved with each other "yours, mine, yours, mine, etc."

This could potentially change the style of play of the game. It may make things less fun. On the other hand, perhaps it could make things more fun (or just 'different fun'). I don't really know.

played a game with the settings

https://www.warlight.net/MultiPlayer?GameID=10627241

lol, are you trolling? ;-) That game ends with a boot before picks are even made. hehe

I like this in the sense that it is a good "proof of concept" (that luck can still be used in strategic settings).

However, I don't see the added value with these "pure luck settings". Is this simply a proof of concept? Or do you think there is an actual advantage to using "pure luck settings"? It's still cool if it's a proof of concept; I'm just curious.

Believe it or not, at least one person has actually argued that most of these players are still newbies, despite being in the top 100 of the 1v1 ladder, which equates to approximately the top 20-25% of ranked, active players on the ladder

You can become a top 50 player by just learning the basics of warlight strategy in my opinion. I'd like to see a few players who have been top 5 in the past play this template before it is considered "strategic". I'll try to find someone to play this with me soon just to test it out.

Sultan finishing relatively poorly is interesting. I think there are 3 possibilities:
1. Sultan just didn't care much about these games and spent very little time on them.
2. Sultan is slower than these other players at adjusting to new templates.
3. This template involves too much luck.

I personally think #1 is the most likely, but it will take a bit more testing from higher level players to know for sure.

As of right now, the template hasn't been proven "strategic" but it has the potential to be proven as such fairly soon. By strategic, I mean the better player should win a majority of the time. It's a vague definition, and for most players, won't really matter. But for the advanced players of the game (like top 25-50 of the ladder), it will. And for this proof-of-concept, I think it should contain less or equal luck than the standard Strategic 1v1 template to be considered Strategic. My bet is it will be equal or a bit higher, but I'll know for sure after playing a few games versus advanced players at it.

However, I was very surprised that no one had even tried the template, and neither had they even bothered to look at any of the 7 example games I had provided, where delay moves were clearly not an issue.

7 example games don't mean anything if they don't involve players who don't play warlight well. Just because delay moves are not an issue in games between basic players doesn't mean they won't be between high level players. Your tournament showed that they aren't an issue in between solid players. At this point, I think you will see high-level players interested in testing it out, as you have shown that this template works with players of a passable level of skill.

In conclusion, cool proof-of concept, and even if this one proves to be a bit too much about luck, I have little doubt it could be on par with many SR templates. The issue is still the annoyance of playing a lot of delays to be on the safe side (there's the possiblity both players play the same amount of cards). However, I think it can be considered "strategic" even if it is a bit annoying. Well done wct, and great idea to create the tournament and show your template works.

Edited 2/28/2016 07:35:48

You can become a top 50 player by just learning the basics of warlight strategy in my opinion. I'd like to see a few players who have been top 5 in the past play this template before it is considered "strategic".

I just don't get this attitude. Is 'strategic' exclusive to only the top 5 players?!?!? Are all the rest of us 'noobs' playing kiddie-warlight? WTF, man.

However, I don't see the added value with these "pure luck settings". Is this simply a proof of concept? Or do you think there is an actual advantage to using "pure luck settings"? It's still cool if it's a proof of concept; I'm just curious.

What is the "actual advantage" to using different kill rates or multi-attack or any other random setting? I don't understand your question.

Sultan finishing relatively poorly is interesting. I think there are 3 possibilities:
1. Sultan just didn't care much about these games and spent very little time on them.
2. Sultan is slower than these other players at adjusting to new templates.
3. This template involves too much luck.

I personally think #1 is the most likely, but it will take a bit more testing from higher level players to know for sure.

Orrrr... for #1 you could just ask him, maybe? For #2 and #3 you could just, I don't know, open up those games and see for yourself, er whatever? Here's the tournament link again: https://www.warlight.net/MultiPlayer/Tournament?ID=16560

Here are Sultan's games:
https://www.warlight.net/MultiPlayer?GameID=10040112 v Ranarius
https://www.warlight.net/MultiPlayer?GameID=10053739 v HyperionKnight
https://www.warlight.net/MultiPlayer?GameID=10064967 v Ender 2010
https://www.warlight.net/MultiPlayer?GameID=10075053 v me
https://www.warlight.net/MultiPlayer?GameID=10079950 v PhucilliJerry
https://www.warlight.net/MultiPlayer?GameID=10094516 v Robert E Lee
https://www.warlight.net/MultiPlayer?GameID=10134826 v John Smith
https://www.warlight.net/MultiPlayer?GameID=10008826 v ntw0n
https://www.warlight.net/MultiPlayer?GameID=10008827 v fireice82

In the game vs me for instance, it's 100% clear that he lost on picks. I knew where he was and went right there. Game lasted 3 turns.

Is that because 1. He was bored, 2. He didn't understand the rules, 3. Too much luck, ........ orrrrr, could it possibly beeeeee, that he lost on picks (which is 'strategy' as far as just about anyone who's not fixated on the top of the top of the top of the ladder is concerned)?

As of right now, the template hasn't been proven "strategic"

By your absurdly over-the-top standards, anyway. Has any template been proven strategic according to your standards? Is the standard "Strategic 1 v 1" template even 'strategic' according to your standards? Cuz mine is basically an almost identical rip-off of that one. So, if we can't agree that the standard "Strategic 1 v 1" template counts as 'strategic' then we're certainly not going to be able to agree that a derivative of it is strategic. So let's get that question out of the way before we proceed any further.

I think it should contain less or equal luck than the standard Strategic 1v1 template to be considered Strategic.

??????? Holy arbitrary, Batman! First of all, the standard "Strategic 1 v 1" template is 0% SR, so how the hell are you going to get 'less luck' than that?!?!? Second of all, why (besides your own personal preferences) must the luck setting be low?

However, I was very surprised that no one had even tried the template, and neither had they even bothered to look at any of the 7 example games I had provided, where delay moves were clearly not an issue.

7 example games don't mean anything if they don't involve players who don't play warlight well.

What part of "neither had they even bothered to look at any of the 7 example games" do you need clarified? How could they know who the players were if they never even looked at those example games? I repeat again for emphasis: They did not look at even a single one of those 7 example games. Not one. Doesn't that strike you as odd? Something that someone might be "surprised" about?

If you really want to test that template, set up a bo9 between 2 very good players and if they don't run into the delay problem I'd be more convinced. However I'm posting for a different reason, one of my proposed solutions to the delay problem was tweaking attack/defense kill rate making it even or making attack kill rate slightly higher like in fast earth. Have you tested one of those setttings?

Strategic 1v1 is 0% (or 16% depending on who you ask) WR not SR. Hence he is saying luck should determine the game less often than it does in a strat 1v1 game with 0% WR for it to not be too luck dependent.

wct if he had picked samerica before greenland picks would have been
you: SEA, Greenland, Canada
him: Echina, Samerica, Indo
your picks were better than his because of 1 slight change in his picks - your picks were not that much superior, he wins.
Even, if he got 1st pick and the picks statyed the same, exactly as what above happens. You lose on picks.

Just pipe down, please...

If you really want to test that template, set up a bo9 between 2 very good players and if they don't run into the delay problem I'd be more convinced.

That's a fair personal standard, IMO. Rather than saying 'it's not proven strategic yet', you're just saying, 'I'd be more convinced if I saw X'. I wish more people would phrase things this way. It's more honest, IMO. It's certainly a lot less mind-boggling.

I think your proposed test is a good one. Unfortunately, I don't qualify as a 'very good player', so I guess it will be up to someone else to try this out. I'm not that great at organizing things myself, TBH. Frankly, I'm surprised the tournament I set up went as smoothly as it did!

However I'm posting for a different reason, one of my proposed solutions to the delay problem was tweaking attack/defense kill rate making it even or making attack kill rate slightly higher like in fast earth. Have you tested one of those setttings?

No, I have not tried those settings. I think you may be right, but I'm not sure. I have a feeling that it would change the style of play quite a bit. The point of my proof-of-concept template was to be as similar to the standard "Strategic 1 v 1" template as I could make it, so that the 'feel' of the game play could be compared against something that's as universally familiar to everybody as "Strategic 1 v 1". I have a feeling (just a hunch), that changing the kill rates like you propose would 'feel' significantly different from "Strategic 1 v 1". That's why I didn't pursue your suggestion when you made it to me in the earlier thread and in our game chat. I would be interested as well to see your suggestion tried out.

Edited 2/28/2016 16:49:37

Strategic 1v1 is 0% (or 16% depending on who you ask) WR not SR. Hence he is saying luck should determine the game less often than it does in a strat 1v1 game with 0% WR for it to not be too luck dependent.

Perhaps the old version of the template used to be WR, but the current one is SR. Go to Create Game, select "Strategic 1 v 1" from the built-in templates. It's 0% SR.

wct if he had picked samerica before greenland picks would have been
you: SEA, Greenland, Canada
him: Echina, Samerica, Indo
your picks were better than his because of 1 slight change in his picks - your picks were not that much superior, he wins.
Even, if he got 1st pick and the picks statyed the same, exactly as what above happens. You lose on picks.

I don't think it's a forgone conclusion that I would have lost on picks. You don't know how I would have played in that scenario. For example, if our picks had been reversed and I got what he got, I may not have invested so heavily in Greenland as Sultan did. That's really what did him in (he didn't lose *just* on picks). I didn't expect him to surrender so quickly. When I saw the history, I saw why.

Just pipe down, please...

NO, YOU!!!1

Edited 2/28/2016 16:59:53

Pretty sure Sultan just lost because he plays like he has two buses to catch in less than 5 minutes. In any case, your template might be strategic (I will not argue around that point), but as JSA said, it's not a very fun template : even if it is potentially strategic, it's not going to be exactly very fun to play with :/. But yeah, if your point was just to prove the strategy of 100% luck, then I guess you did

I think it's great fun personally... But then I suppose players who like the current 1v1 ladder aren't going to like it...

Pretty sure Sultan just lost because he plays like he has two buses to catch in less than 5 minutes. In any case, your template might be strategic (I will not argue around that point), but as JSA said, it's not a very fun template : even if it is potentially strategic, it's not going to be exactly very fun to play with :/.

This is more speculation that doesn't stand up to scrutiny. If it's not fun, it will be not fun for reasons *other* than the Delaymageddon scenario you guys are imagining.