Opening Tree

Opening treeReview details:                                                                           
App version:  1.2
Device:  Ipod Touch 4th ed
Operating system: ios 6.1.6

Summary

A potentially useful tool for learning about chess openings but rather limited and let down by a small database. Design and layout makes use on a mobile device difficult.

Opening Tree is a free app from Michael Adams who is responsible for a range of chess related apps and software. It is also available for the Ipad and includes an inapp purchase giving the user added features, in this case the ability to edit, score moves and make notes. This review is of the free version (ie without the additional purchase option).

Features

Opening Tree is an analysis tool which enables you to study chess openings in two ways, using either:

  1. information from a database of opening moves which itself is drawn from the results of a large pool of games; or
  2. a chess engine to provide detailed analysis of specific positions

Opening tree_3The database of moves with associated frequencies of wins/draws/losses is very helpful and is a commonly used format in established chess database applications. It enables the user to quickly see which are the most frequent moves from a given position and also which produces the best results. As a small improvement, in addition to W/D/L information the % score for each move would also be helpful to more clearly identify the relative success of each move.

The quality of the database of opening moves used is crucial for this to be a useful tool for analysis. In this regard the Opening Tree app is somewhat disappointing. The size of the database used is relatively small, (due to memory constraints on the ipod?). This means that it doesn’t take very many moves, even in the more popular openings before there are either no more or too few results to be meaningful. As an example, if you want to study the Budapest Gambit (1 d4 Nf6 2 c4 e5) you’ll find the database has the results of just 33 games after only the 2nd move. The other drawback is that there is no information included within the app to help assess the quality of the games used to build the database. For example, games where both players are rated above say 2500 elo might be considered a better reference source than where the minimum rating was 2000 elo.

The inclusion of a chess engine to provide analysis of potential moves is Opening tree analysisalso a good feature. This enables the user to check and assess possible opening lines and counter moves. The chess engine included with the app is Crafty 23.4 – a longstanding and well-respected engine, which happens to be one of my favourites. Whilst Crafty is certainly strong enough for this purpose for the average app user, I can’t help thinking that most knowledgeable users would prefer one or more of the top strength engines that are freely available (for example Stockfish) for analysis purposes. Sadly, Crafty is not one of these.

There are additional analyis features. An in app purchase allows you to assess and record particular moves by using different colours to signify good and bad moves and you are also able to make notes about speciifc moves. The review didn’t test this functionality.

However, it is after a while of using the app that some of the weaknesses with the concept begin to become apparent. Yes, it provides a good way of reviewing and assessing opening lines but it is rather limited in how this can be achieved. The app has a lack of accompanying features that would be helpful for studying purposes. For example, the app doesn’t actually let you play or practice any opening  – the engine is only available as an analysis tool and not a playing partner.

Learning and user edited work on opening lines can’t be exported for use in other apps, nor can the learning be utilised in any other practical way by the app. For example, the app doesn’t allow the user to generate an opening book from their research to use either in the app or another alternative program or app. So there is no direct way to use the app to play against or test your own research and pet lines, you will need to fire up and another chess app.

The user can import their own PGN files of games into the app either via email or ITunes file sharing. However, due to technical constraints (essentially catering for the needs of older devices) only small files of no more than 500 games can be handled. On the Ipod once a PGN file is loaded it is a terribly fiddly and frustrating job to ‘step’ through the moves of a game to be able to compare them to the database. This is down to the size of the controls more of which below. I found myself veering between stepping forward and backwards too often for comfort.

In summary, in considering the combination of features and flexibility I was left with the feeling that a dedicated chess database app might not be a better option.

Practical use and presentation

Opening tree_mainThe first thing that’s obvious when using this app on a mobile device is how small the chessboard is. The board takes up just over half of the screen size and means that not only is it potentially difficult to view but that the dreaded ‘fat finger’ problem is likely to come into play. The result is that if you want to enter moves manually (rather than via tapping the particular move from the movelist) the size of the individual squares makes it tricky to tap the correct piece. There feels to be areas of wasted space on either side of the board (maybe the arrow keys could be located here?) and there is no option for resizing the board. Neither is there any potential for screen rotation, and I wonder if landscape format would be better at least on an ipod (this is available on the iPad version).

These issues combine to leave a feeling that the design of the screen layout for the ipod/iphone user at least, could be better.Opening tree_options

However, although size and layout can’t be altered, there are opportunities for tailoring the presentation. Within the ‘Settings’ option there are a range of easily accessible options for changing various visual aspects. These include a good range of different chessboard colours, background colours and at least two chess piece sets. Refreshingly (unlike some other apps) the choices are between well-designed and respected piece sets which are both very clear and comfortable to use.

Developer support

The app description on the ITunes store is failry descriptive in terms of the app’s features and operation. This is supplemented by some specific notes of instruction on selecting relevant options; in particular when loading a pgn file for the first time. The developer has also included a change-log on his website together with contact details for support.

To date the app has had only one update since its original release, this being a year since the writing of this review. As such it doesn’t appear to be in active development. However, the author has released and/or updated other chess apps in the intervening period, so it is too early to conclude that the app is dead.

Overall


Likes:

  • Free and ad-free
  • Inclusion of chess engine for self-analysis
  • Good choice of visual options
  • Simple and easy to use


Dislikes:

  • Chessboard is too small to view/use comfortably
  • Small size of database of opening moves
  • Uses a relatively ‘weak’ chess engine
  • No recent updates

 

These are my thoughts; if you have used this app what do you think…..?

14 thoughts on “Opening Tree

  1. I”m a little late to the party it seems. Hi i’m the developer of OpeningTree. The smaller board on iPod still exists. iPad is great. iPhone screens have gotten bigger though and I think the issue is much less on larger size iPhone screens and have tried it on some. Many updates last summer. The in app purchase of scoring moves is now free. PGN loading really got the most work. Now 2500 games not 500 but also other nice enhancements. Two key speed metrics have gone from maybe slow to instant. Files opened are cached during run time life of the program so they load instantly. There was latency tapping on a game and it loading on board. This is now fixed and instant. Bugs in reading PGN files fixed. We read now all files and games i tested were before some broke it. On first opening PGN the app still explains how to get PGN but has an option to install a collection of 6 PGN databases that include Kasparov games, Fischer’s memorable games, two recent international tournaments that have games from those like Carlsen and Nakamura and classic games. PGN presentation is a little nicer as well with players names at top of board now. We also finally replaced the sort of shoddy icon with a nice flashy new one an artist friend created. the book is still small but due to database building improvements developed as i ported the book to other apps of mine, it’s possible to build a new one now that is several times larger yet smaller in file size than the original book. So that may be an update. The games are actually high rated. The lowest I went was both ratings had to be over 2300. I think half are over 2400 both players. Use is up by stats Apple reports ( only a third opt in for this reporting apple tells me) from 10-15 users a day ( multiply by three is supposedly the full number) to 25 now ( again multiply by three) and in total nearly 1000 a month estimated. I did like this review because it’s a good assessment of the app in ways though some things have changed like bigger phone size and bigger PGN files that can be loaded.

    Now on the stockfish issue. Android can load another program. IOS apps cant. To have Crafty the Crafty engine is compiled in OpeningTree’s code. By compiling directly with engine code you link directly to that code and as such as subject to any license related to that. Stockfish is GPL and GPL requires that if you link to it you must be GPL in you’re code. I’ve participated in some discussion with people knowledgeable of the Stockfish team and it sounds like they might let that slide i.e. if you don’t change Stockfish we won’t bother you if you say you’re using Stockfish. But having a GPL program myself, i’m just still not comfortable with that as a legal basis to use Stockfish.

    Like

  2. OpeningTree 1.60 is released. database went from 60,000 games to 200,000. All over 2300. Now at d4 nf6 c4 e5 there are i think 113 games not 33. Move not button introduced. Engine play possible. We stop the engine analysis on move now. Move made; move is what we say.

    Save board to PGN possible. Player names are taking from opening name for the PGN file. The file can be mailed out and exported.

    The small board was an iPhone 4 issue. all my apps use width of board equal to height. Openings was optimized for the iPhone 5 tall screen height.

    Like

  3. Many thanks Michael for taking the time to respond to this review and for your detailed summary of changes you’ve made to the app since the original review. It is great to see that it has been developed further. Thanks for continuing to make enhancements and good luck with any future developments

    Like

    • Many thanks to you. OpeningTree was one of three chess apps made when I took a one year sabbatical in 2014 to develop iOS apps to transition to being a mobile developer which I now am. It was not one update when you saw it but 2. 1.0 1.1 1.2. the change log did not mention 1.0.

      So I wrote high rated games. OpeningTree was a shot in the dark at a successful chess app. But a friend liked it. The first version only had the book. He said it be nice to mail some games and check them against the book. I had pgn parsing code I used to make the book. So I did load PGN. And a bit later I added crafty in 1.2 with Bob Hyatts personal permission with the condition I say i’m using Crafty.

      It was over 2300. maybe a few hundred in some special collection I did for a friend went to 2000. Some stonewall position. Hard to find those. I notice we have no King gambits now. I guess both over 2300 since 2010 no longer play the kings gambit. I may boost that. The first book, 60,000 games, was not strictly proportional in opening move frequency in the early stages. I boosted Eco’s loading in files like D00-D99. E00-E99 and even some sub categroies like just A00. Half the games were general over 2300 used to seed the book.

      The latest databasing was time consuming. Files that started out taking 30 minutes grew to an hour over time. I had roughly 26 files I had split of 5000 games. And my mac mini over heats so I had to shut down 30 minutes after each file. I spent my time on my Android tablet using the Android version of Lantern Chess, the Nexus 9.

      It took maybe 40 hours. I want to grow to 250 maybe 300K games but the technology will have to be looked at. I wrote my own book builder. The project was scoped in my sabatical to both write opening tree and the book builder. Maybe it’s database latency, maybe its my pgn parsing which as we know has been slow in openingtree.

      OpeningTree is the only chess app my dad has ever used and liked. Over Christmas one year in Reno he spent the weekend with it. He liked he could as he put it “help the players”, he was using Crafty in conjunction with a pgn file from a tournament I mailed to his device.

      Like

  4. One last comment for clarity on OpeningTree’s game collection which i mentioned is both players rated over 2300. I’ve said to friends i wont specifically say where i got the games because i don’t want a competitor to say hey you cant use our games in product that competes against our products coming at me. But I told the friend I see no copyright issues as my book is an abstraction of the data set and its own unique thing not a mapping from anyone else’s data to my data anyone could do. Also commercial databases do bill that they are for professional use. But the games i used are over the board games and they are also games I paid for not an internet download. I felt the quality would be higher in a paid collection than what you find in downloads.

    Like

    • Sorry i was re-reading what i said about us having no king gambit games. I mean there are very few. After e4 e5 f4 there are only 16 games which seems incredibly low now that we are at 200K games databased in. But it may make sense if it means that when two 2300s or higher meet in a game the kings gambit is not much played post 2010. After e4 e6 d4 d5 nc3, a line in the french, there are over 4000 games and after e4 e6 d4 d5 Nd2, another popular french line, there are over 2500 games. So i think openingtree’s book reflects a lot of what is played in modern games by two masters. I may boost the kings gambit though. Not by a tone to throw the distribution off to much but enough to make openingtree usable to study it.

      Like

      • I dont want to boost d4 nf6 c4 e5. We are at 113 games i think now and its a sort of ok sample size if you consider with both games over 2300 results of say 5 games are meaningful. But i’d like to get up to 300K or 400K games databased over the next few months. Even with current technology its a bit time consuming but i can schedule it in over a longer time span to be a bit less tedious and not done in 5 days. i got 150K in in 5 days i can in a few months i think get to 300 or 400 total. And this will keep me honest and help everyone in a similar situation. But the kings gambit i suspect needs a different treatment as its not a modern opening anymore.

        Like

  5. At the risk of over extending my welcome for replies i have to say about this comment of yours what has been on my mind. “games where both players are rated above say 2500 elo might be considered a better reference source than where the minimum rating was 2000 elo.”

    now i realize you are painting a spectrum there and not disqualifying both players over 2300. The idea for both players over 2300 came from the original fritz power books in the early 2000s. though i think they did eventually at a higher price do a 2400 book. 2300 had an appeal to me as sensible. In general terms it means both players are fide master or higher. It is a good definition of master games. Now with commercial databases i dont know the frequency they use FIDE ratings over say national like USCF. So it may not be just fide masters and up. But its what i’m aiming for.

    Like

  6. Now no professional chess engine competing in the world chess championship uses anything like a power book despite chess base selling this. They have custom book builder that tailor the book to the engines strengths. Just because game databases say some moves are played doesn’t mean you want to play them ever. but this is more important the first few moves for a chess engine and it can i suspect use the high rated winning responses a bit later in the tree. But i felt for an opening book for the general public there was no point in tailoring strictly on a power objective. Some play the french. They need the best responses even if a custom book builder decides it doesn’t want fritz to play the french.

    Like

  7. I”m working on the OpeningTree 1.70 update. Book is going from 200K games to 300K. I’m going to take my time more and not do the databasing in 5 days ( then 140K games) and take ten days for 100K games. 13,500 were databased in today. The new games are like all over 2300 from 2017, then 2016, then 2015, then 2014. We are on 2014 now and will get all of 2013 and probably some 2012. The old book did not go beyond 2011.

    I did boost the kings gambit by 300 games today and was pleased with that. I was wrong it was not 16 games. I must have had my move list scrolled down. There were 170 games. Budapest had 113 so the kings gambit is maybe 50% more played. After seeing how well it went I decided to boost the Budapest and boosted it by 200 games. Both searches were 1995-1999 both over 2300 position search. I wanted to use some back years but not to far back to reserve recent years for chronological building so I don’t get duplicates.

    I have only boosted in two ways, position search and eco file. So far only 3 positions have been entered in by position search. A stonewall, the kings gambit, and the Budapest. To not disturb move frequency i’m only doing these very early in move order like move 2 or 3. Eco boosting was done in the original book after the first 30,000 games were databased in. There were files by general eco like C00-C99 and some more specific such as the Ruy Lopez and Sicilian. I don’t feel opening tree has to be totally reliable on move frequency on the first 3 moves but after that I have not boosted so the frequencies once in an opening are sound.

    Like

  8. I decided that instead of doing 300K games and another release I’ll database all the way through the 2012 games for around 130k and release at 330k. The conventional database in the first release was 2008-2011 games. That’s games from last decade, and I don’t want to go back further as I want a modern database. I am willing to database in future years and boost less played openings upon request with games pre 2008.

    I have boosted the kings gambit with 95-99 games and i’ve boosted the Budapest with 95-04 games about 384. I’m going to add in the 05-07 games as well since that takes it up to when the conventional database starts. It will be pretty much all master Budapest games played in last 20 years. I’m guessing 750 games. It’s ranked already a bit down in move list and I don’t think move frequency issues on a low ranked move 2 are going to matter much. The win percentage for e5 has dropped from .43 to .40 and is staying constant. It’s hard to win with in master play. Losses outnumber wins. But its got a lot of draws too so many masters may feel safe they know how to draw. In club play many don’t know it as well so it has surprise value I would think.

    I feel while fewer games than some competitors that boast a million or two, the quality is high enough that down to 5 games for a move its more trustworthy. That means 5 masters saw the move. Even a final move with 3 games and one move could be trustworthy as it means 3 masters seem to see it as forced here.

    Like

  9. OpeningTree 1.70 released today. I decided to take games from 200,000 to 240,000. The time to do more was to much for my schedule at the moment. But there is now in the App an About OpeningTree menu item on Actions menu where I explain how the book is built, game quality etc, years covered and offer for those who contact support to boost less played openings(and the criteria such as opening entered into early in move number) with games before 2008, when the conventional database starts. I have in 1.70 boosted the Budapest, d4 nf6 c4 e5, and the Kings Gambit, and this is mentioned in the About OpeningTree in App.

    Like

  10. Did two updates 1.71 and 1.72 to fix bugs. That should be it for now. Key in 1.72 was database latency was fixed. On most devices i tested on this wasn’t a big deal but on my old iPod touch 5th gen it was taking 2-3 seconds to respond to a move tap. Now down to 1/2 a second to a second. This should also make databasing new games in go faster when i get back to that. In 1.71 one bug fix germane to this article is when d4 nf6 c4 e5 was played the user has to scroll down to e5. Before 1.71 when new moves loaded the user remained scrolled down and the lower down the scroll view moves of course have lower numbers so it may look like there are no games. It now auto scrolls to top on loading new moves. I think this is why i thought the kings gambit had no games. it had the same issue.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s