Dominion Strategy Forum

Please login or register.

Login with username, password and session length

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - SettingFraming

Filter to certain boards:

Pages: [1] 2 3 ... 6
Dominion Online at Shuffle iT / Dominion Scavenger is Back
« on: December 24, 2018, 11:36:56 am »
The online rankings site Dominion Scavenger ( has been down for a little over a month now, but I've finally had the time to make the changes necessary to bring the site back up.

Basically I cache a lot more aggressively now, so instead of constantly trying to pull latest results from the live Dominion Online servers, I pull them every minute and cache everything so when individual people load pages it all just happens on my server and doesn't reach out to Dominion Online's servers, potentially overwhelming them. So results should be updated every minute on Scavenger now (both on the "Scavenger" page as well as the "Live Leaderboards" page). As a side note, I've expanded the live leaderboard page to be the top 500 instead of just the top 200 now ( - must be in top 550 to start the day to be eligible).

My life's been too busy the past month to work on any of the suggestions that I've gotten for the site (quite obviously, as I didn't even fix the site), but the next few things I will work on will likely be:
  • Fixing the mu/phi/levels graph
  • "Recent live leaderboards" - such as leaderboard based primarily/only on the last 30 days of results.

I've also had several offers of people offering their help to work on the site. While that isn't currently too easy (my dev environments/repository settings are not easy to replicate as of right now, and most people aren't Django developers, which is what the site is written in), I want to make that more possible than it is right now, so I have a goal to make that better.

Hey folks, I'm sure most of you already know about Dominion Scavenger, the live rankings and leaderboards site that I run as a companion to Dominion Online.

The site has been up for about a year and a half now, and now that I've got some time opening up again I'm looking to add some new features to it. I want to know what to prioritize based on what people would actually use. The core functionality of the site will always be being able to see your rankings change after each game, but there's a lot more already to the site and a lot more I'd like to add.

Things I'm going to add soon to the site:

  • Making the dates for the past levels/mu/phi chart on the bottom of more configurable.
  • A little data dictionary explaining some general guidelines for how the rankings work

Ideas I already have or have been given for the site, let me know if you support wanting any of these:

  • A "Recent Past" leaderboard showing who has been the best player not just of all time, but also over the last 7 days and 30 days (this would use glicko2, but a slightly different formula to reflect the condensed time period).
  • Ability to download various things such as past results as Excel natively
  • Ability to view historical leaderboards easily (i.e. rankings as of December 31, 2017, etc.).
  • A forecast tool telling you how different hypothetical results would affect your level
  • Any other good suggestions I get!

Things that aren't all that possible:
  • Your past record against a single player (requires some database stuff that I'm not able to do given current constraints).
  • Features that involve game logs for every game, as the game logs are not stored in a terribly parse-able format.

I also have some vague plans to add other larger sections to the site, such as links to other Dominion content, a blog section featuring some specific good articles (or, alternatively, articles I myself have written), and/or various other Dominion data sections.

Please let me know what would be of value and any other suggestions you have, I want to make the site a great place to view Dominion stuff between games!

Dominion Articles / Re: Shepherd
« on: February 21, 2018, 09:07:20 am »
This article contains a lot of incorrect information, in my experience. Shepherd can be made to work a lot of the time in a wide variety of decks, the points that going heavy Shepherd supplies are kind of a big deal not infrequently, what aku said is absolutely critical to include in a Shepherd article (as well as the possibility of purposefully triggering terrible green-filled shuffles with i.e. a saved Shepherd), and the math about Shepherd density is very misleading with respect to how to actually play the early game.

I appreciate the article, but I do have a ton of points of contention with this as to where I feel that someone who has not yet played with Shepherd much will walk away a worse player than they were before if they try to stick to this article's principles.

Dominion Online at Shuffle iT / Re: Is it okay to buy Possession?
« on: November 08, 2017, 01:16:35 pm »
Possession being banned from randoms is great, for two fundamental reasons:

1) The common play when Possession is good is to build a deck that can only play Possessions, and then you slowly squeeze down piles while playing multiple Possessions per turn in deck that has either none or nearly no economy. These games are not interesting, after you've played one of them, and perhaps more crucially they take about an hour and a half.

2) There are many Possession interactions which cease to be in the spirit the card was intended to be used in, and these are decidedly unfun. A lot of these are Adventures/Empires related, as reserve mats (especially with Overlord/BoM!), token placements, debt, etc., are just broken interactions. Also, cards like Masquerade and Ambassador are interactions that cause Possession to be devastatingly annoying to play with.

There is a third aspect that people don't like about Possession, which is the unfun nature of having someone play with your deck. I don't think this is ban-worthy (at least in the global sense), but I certainly get how annoying it is. The worse part of this, and I think this is something that strikes more at the heart of something I really truly dislike about Possession always, is that suddenly it makes turn ordering matter a lot more. By this I mean when you're playing Dominion decks and you dud it usually means your next turn will be good since you just got your dud cards out of the way. But no such thing exists if you get possessed on your good hand, and this can be incredibly frustrating to have happen to you.

And this isn't even mentioning all the stuff I'm forgetting, such as the pins and semi-pins such as bad shuffle triggering.

Tournaments and Events / Re: World Cup 2018 Interest Thread & Discussion
« on: November 04, 2017, 09:38:15 am »
I would like to join, but I don't think there's any other danish players.
There is a danish facebook board gaming group, where I can try to ask and search for 3 other players (or however many is required).

Emil is Danish

Like 3-player! ;)

3-player is actually the next thing I plan to do at this point for the site. I wish there was a noble reason, but really it's just been that I've actually recently caught myself enjoying 3-player Dominion.

No idea when I will finish it, because I haven't started. It's not that hard to do on the backend because the database stores results in such a way that I should basically just be able re-use all the two player code, but making the interface easy to toggle between 2-and-3 player will be the trickier part.

Quote from: Cave-o-sapien
This seems like the key phrase here. It seems like it should be possible to mine this sort of historical meta data into a separate structure that IS optimized for such queries. Obviously this isn't something YOU can/should do, but it would be a cool thing to have.

That's right, it's theoretically possible to partition/index/build to make lookups like this fast, but it isn't really practical to do so in this case. You would have to either build another index against the table or use a data store that's more optimized for these sorts of look ups, and that's too much work to make something like this possible (as well as, as you mention, largely out of my hands).

I hate the cards that play terribly like Page.

I like the cards that play really well like pretty much all of them.

The overall design of Dominion is wonderful--I don't really hate anything there, and any exacerbation of luck, etc., has more to do with specific cards than the game design itself. And the fact that I only really truly dislike about 4 cards is saying something for a game of this scale.

New request: would it be possible to find query head-to-head matchups between any two users?

I actually implemented this a little while ago, but it's too intensive on the database because it has to search back through all of the results indefinitely in order to locate the results between two players and the database just isn't built for that (there are a lot of results! Evidently gokosalvager had a similar thing at one point, and it was removed for similar reasons.). If I do a "compare players" feature it would just be something that gives X win%'s and the like, but that's not too high on my priority list of things to do.

How do you grab the logs manually?

Copy-Paste from in-game. Unfortunately logs go away as soon as game ends now, so you'll have to re-load games off the game ID's if you aren't able to copy the log before the game is over.

It'd be much better if the log still stuck around after the game ended, so you could copy it much more easily.

The Dominion Scavenger website now features a way to save logs and view them in a somewhat pretty format.

You still have to go manually grab the logs at first, but you can paste them on this page here:
to create a pretty log.

That will create a saved, shareable link where you can then post a link to the prettified version of the log. Here's an example of that:

This is Version 0.1 for many reasons, but mostly I'm just releasing what I have right now since it makes for a convenient way to save and share gamelogs, and at least the logs are much easier to read than the text blocks we're currently having to work with.

Known things, for which the reasons are either technical or I just haven't had time to get around to them yet:
1) The colors are a little off (Curses, for example, are blue).
2) Not all colors are there (Reserves, Ruins, Shelters, multi-type cards), etc.
3) There aren't any of the context features which would be nice to have (Kingdom listed at top, deck trackers, etc.).

It's still a far cry from the feature set of Salvager, but hey, it's something!

Thank you, everyone, for your kindness.

This is a fantastic resource that I use with almost every session. Much appreciated.

A couple questions:

1) It seems that the Y axis labels on the ranking over time chart is rounding in an unexpected way.

2) Is there a way to see the ranking over time over a longer period?

1. Yes it was rounding a bit too aggressively (and it was truncating instead of rounding). This has been fixed now.
2. I changed it to 60 days now instead of 14. Ideally it would be configurable, and there's no reason it isn't except for the fact that I haven't done it.

I use this website all the time to see where I'm going to be after the next leaderboard update.

I think it would be helpful to have your current rank on the learderboard displayed on your page.  I know this is available elsewhere, but if you want all the ranking information in one place, that would be included.

Yeah that would be nice--it wouldn't be too hard to put the daily leaderboard position on your page. The live leaderboard would be a little tougher since (A) it's only for the top 200 right now, and (B) I have to load a lot of stuff for the live leaderboard updates so I'd have to be smart about how I do that to avoid too much server traffic.

Thanks a lot for making this tool!
Just fyi, it looks like something went wrong with the leaderboard today and a lot of players are missing.

Edit: Seems this is fixed with today's leaderboard update. Must have been a fluke.

I cache those leaderboards every day, and sometimes a partial copy gets cached. I've done some things to try to mitigate this issue, but it shows up from time to time. Usually if someone pings me on discord it's easy enough for me to clear the server cache and fix it. I'm way more active on the Discord than I am on this forum.

Dominion General Discussion / Re: The swingiest card ever
« on: July 06, 2017, 08:18:01 am »
*cough Black Market *cough

Nope, Black market isn't swingy at all, but there are some people who like to blame their losses on this card.

I patently disagree with this. What and when you flip from the BM deck is incredibly important AND not in the player's control. Thus it's plenty swingy and I have won and lost plenty when the most important meter movers in the game were Black Market plays.

Page, though, receives my vote.

Dominion Articles / Re: Refreshing the Dominion Paradigms
« on: June 06, 2017, 06:01:10 pm »
Thanks for the explanation. Makes a lot of sense. The difference I described would indeed arise through different evaluations of tempo concerns. And it is fair to argue that this is more of an in game tactical insight then it is a strategical difference of deck type.

So the open question is then if there are deck archetypes that don't naturally arise out of your 4 point engine strategy. (and are not BM, slog or rush). (and can be optimal on a reasonable amount of boards). I wonder what ideas people have about this.

My engine categorizations are pretty broad, so it's probably almost always going to be possible to categorize every engine in to one of the four (if, sometimes, the line being blurred on which category a deck belongs to). I could certainly see people dreaming up things that don't fit at all though.

The main point of this article, really, was to try to introduce focus into engine play. Rather than just thinking "I'm going engine", to think more along the lines of, "I'm going engine, and this is what I anticipate this deck being capable of compared to the alternatives, and this is how I'm going to play this engine differently than that engine".

Many experienced top players don't usually need to think along these lines as they can see a board and digest it on a more natural basis, but I think there's a lot to be learned by players who struggle to i.e. make big explosive decks work, or to choose between money strategies and weak engines, etc. And even top players can struggle with these things from time to time.

New feature up: Live Leaderboard.

It's updated instantly throughout the day, and available at:
Due to performance concerns I've limited it to only the top 100 players, so it's really just for watching the top of the race. I may make a live leaderboard that contains more players, but it would require me to throttle/cache the results.

Leaderboards are now available at:

The in-game client has had some trouble updating the leaderboard display the past few days, but this display is up to date.

The default is a leaderboard of the top 1,000 players; there is also a full leaderboard option which has everyone that has ever played a ranked game on the client.

Feature request: keep track of win rates as 1st and 2nd player, respectively.

I don't have access to this information (the only way would be logs, and logs are not machine-readable by anything except the client itself right now), so this won't be happening until if/when I do.

Puzzles and Challenges / Re: 3 pile practice
« on: April 19, 2017, 11:07:13 am »
Here is a very difficult one. You start the turn Enchantressed, but you will reveal Stonemason from the Black Market part-way through your turn. Lots of Procession going on here. I failed on my first attempt at this (was $2 short), and I'm not even sure if it's possible, but I have to think that it is.

#2928750, play as me, the decision index should be set up correctly to where you can just go.

Is there a way to easily see your all-time W/L record?
Also a full leaderboard list would be interesting.

I'm going to add expected/actual W/L to the "Get Last Results" screen soon, so you can see it there by just typing in a high number. It's currently limited to 1,000 results, but I might eventually bump that up. You can expect this pretty soon.
EDIT: This has been done.

Full leaderboard is on my list of things to do. I just need to cache the results so I'm not hitting the DB for everything each time there is a request to see it. Not sure when you can expect this.

New feature up: chart of your level changes over time now appears on the Glicko estimate page.

Feature request: being able to type the username directly into the URL so that I can bookmark my own results.

This is done. Once you've navigated somewhere you can copy the URL and then all the things that you would expect to work should.

Why is the title of the page "Dominion Sparkle"?

Great work by the way, this is super useful.

I've asked for naming suggestions of several different things over the past few months, and sparkle has been involved a lot. As for this website, maybe I should just call it Dominion Scavenger instead. So hey, maybe next time I edit something it might be that. Glad you like the tool!

"X Win %" is using your current level, right? So you see now the "real" expected win percentage, not what Glicko-2 thought then. (Else, all games on day 1 would need to have 50%.)

That's right, and it's calculated based upon your current mu only (rating deviation is not used for implied win%, which is consistent with the internals of Glicko-2). While it's technically possible to go back and retrieve what your Expected Win% was on the day of each match played, it is (A) computationally a lot more expensive/difficult for me to program, and (B) arguably not as useful and nice to look at as current level comparisons anyways.

First things first just go play around, as that's better than my description will be: Dominion Scavenger

Stef was kind enough to grant me database access in order to make my Ratings Estimator much more accurate, and also not make you guys keep track of your own results. So I built a tool that has a bunch of features, and should be way more useful than that ratings estimator thing.

Live Glicko Rankings

Live Results History

1) Live Ratings. As soon as you complete games, you can click "Get Glicko Estimate" and receive info about what your ranking would be if the period ended now.
2) Opponent information and Expected Win% information. You can see your implied chances of beating any given player, as well as detailed information about players you've played today.
3) Games History. You can view past results, with all of the above opponent information included.
4) A graph of your level changes over time.

Features to come, probably:
1) A graph of your level change over time.
2) A "compare players" feature.
3) Game Id's included so you can find old games to load up and replay.
4) A top "X" (100?) leaderboard that's updated throughout the day.
5) Make tables sortable.

6) Other stuff that people point out to me as useful.

Dominion Online at Shuffle iT / Live Ratings Estimator
« on: April 05, 2017, 03:53:43 pm »
Edit: You should probably just go here now:

If you're as fraught with grief as I am, then daily updates of the leaderboard at 00:00 UTC might not be enough for you. Or maybe you just want to receive numerical confirmation of your ability as a human more frequently than that? Well, regardless of the situation, I've built a tool to estimate your rankings at any time based off of your current ranking and the players you've played against.

You just have to keep track of the level of players that you've played and your results against them. Then you can go here: and enter your current ratings and those results and receive an estimate of what your new level will be if the rankings were to update right then and there.

New Feature: If you don't keep track of every game, you can just enter in an estimated number of wins and losses as well as your estimated average opponent's level, and still get an estimate.

The number won't be 100% perfect because in order to keep things simple you just enter your opponent's level and I estimate their mu/rating deviation from their level, but the results you get should be really close to what your actual update should be.

Again, that's

Dominion Articles / Re: Refreshing the Dominion Paradigms
« on: April 04, 2017, 02:21:49 pm »
    However, your second advice is severely misleading: Cards, events or landmarks that give points are usually a major reason to go engine, even the as-good-as-it-gets-version! This is very important to understand, that's why I posted it. Your argument

     Buying a single Ill-Gotten-Gains and otherwise playing straight big money beats straight Big Money 70% of the time![/li][/list]

    refers to straight BM, which is usually an irrelevant comparison. Nearly always, alt-VP favor the engine.

    Sorry if this was misleading, but that's not what I meant to imply at all. You'll notice that the quote you've pulled is from the section on how to play money decks, not in the section on when to play non-forced money. So my point still stands: points (even just one) are really good for big money once you've already decided to play big money, especially in a mirror.

    I don't disagree that the presence of Alt-VP tends to make Engines better, of course, but "Curse" is obviously not Alt-VP.

    As for your first point on kingdom treasures, I must grant you that perhaps I've over-stated their general importance.

    Dominion Articles / Re: Refreshing the Dominion Paradigms
    « on: April 03, 2017, 05:43:46 pm »
    Many of us remember the terminology established by Wandering Winder on deck types: Engines, Big Money, Slogs, Rushes, and Combos. I would argue, however, that thinking about decks in these terms has become increasingly irrelevant

    And I would argue that it has become increasingly relevant. The fundamental deck types aren't a spectrum, they are very distinct from one another, and the existence of each type of strategy relies on particular rules and card interactions that are unique to that specific type of strategy. With new expansions, it's possible that new types of strategies pop up (such as the coin token hoarding strategy, which is technically a thing, it just sucks in practice because the coin token cards aren't strong enough for that to ever beat anything) and I wouldn't even be surprised if there turned out to be undiscovered types of strategies that are possible with the current cards, but the principles that the original 4 main strategies were built on do still exist.

    They still matter, of course, but my main argument is that with the vast majority of games being engine games nowadays, knowing the distinct deck types in the same manner as before is not as helpful as it once was when you more often played those non-engine distinct deck types.

    Your engine classifications are, however, just arbitrary points on two spectra: how reliable the deck is and how big the payload is. For all engines, it is universally true that first you should thin as much as possible, keep a little economy around so that you can actually build the deck, start adding more engine parts, and when you're drawing reliably, you should add as much payload as you can considering that it makes your deck less reliable and that while you're spending time getting it, your opponent might be spending his time getting points and ending the game. If that means you're able to buy 8 Provinces on the turn when you finally start to green or that you're able to thin very quickly to build a very reliable deck or that you can't build a reliable deck at all or that you can build a decently reliable engine that doesn't have very amazing payload, none of it is particularly noteworthy, that's just how things sometimes work out naturally as a consequence of playing the strategy the exact same way every game.

    I'll grant that for the most part the engine classifications above are generally based upon the criteria of reliability and payload. However, I do make the (and I know it's relatively weak) claim that you should play engines differently based upon those criteria. You mention that it's just "what ends up happening", which, while true, there are definitely things you can do if you properly anticipate "what ends up happening" that can give you an edge. For example, in the above no-draw Highway stack perhaps you should buy Gold before greening, or else you're not going to be able to buy anything useful after your first Provinces. These engines build differently, even if they do share the same core principles. Whether or not they build differently enough to deserve to be treated sub-categorically as different, well, that is not for me to decide; only to suggest.

    Also, the fact that engines are able to play a bunch of Action cards every turn is neither useful nor a distinction from other types of strategies: there is no inherent strategical reason to care whether or not it says "Action" on the card, and some big money, rush and combo decks can also play a bunch of Action cards every turn.

    You're missing the point here. Playing three Ironworks is not "playing a bunch of actions". Playing two Poachers and a Swamp Hag is not "playing a bunch of actions" Playing 8 Merchants and Goons is playing a bunch of actions. I didn't want to put a number on it, but hey, I'll do it. If you're playing on average 4+ actions a turn, it's an engine. Otherwise it's not. So it's definitely distinct, and I very much think it's useful because, hey, you should really know what your turns are going to look like. Of course there are shortcuts in any lexicon and formalization of a definition, but I would argue that this is completely sufficient.

    Pages: [1] 2 3 ... 6

    Page created in 0.061 seconds with 18 queries.