My greatest misses: overestimating getting blitzed

 In this match against XG Mobile I made a horrible mistake, and one that took me a long time to even understand.

White to play 41.

So what do you think? Play it safe or split the back checkers and try to run?

My reasoning was: Black has 11 men in the zone, if I split now with 23/18, he will build a big board or prime and just kill me. Better play it safe and wait for a good roll to bring my back men home. So I played 9/8, 9/5 and was very proud of this move that gave me much flexibility in the next rolls.

XG2 said: -0.46. Biggest blunder in weeks, recommending 23/18. 

That really was a WTF moment for me - splitting my back checkers against so many of his builders? Running into a 4-way attack? The only numbers that don't hit me are 55 and 25. 

Only once I played 10 games from this position I noticed the difference: black is closing in fast. If I don't act now it will build a strong prime in the next moves. But if I split and endure his hit, I have some good fighting chances because of its weaker inner board. Many hits give me return hits.

The lesson this taught me: don't overestimate the opponent's blitzing chances if their board is weaker than yours. Is that the lesson? Hm.

Watch them UBC videos!

Hi everybody,

If you have plenty of time to spare (supposedly there's a certain virus that gives us all kinds of free time that we formerly spent partying and such) and want to improve your backgammon, then have a look at the UBC (ultimate backgammon championship) videos on youtube.

These videos are well-produced, feature great commentary by backgammon grandmasters + analysis by XG; the games are very dramatic at times, and best of all: this is a "backgammon galaxy" style format: Players play several matches to 7 points, and get a point for winning a match, and a point for having played with the lower PR. This way the players spend much time thinking about the best move, and the grandmasters commenting spend this time discussing why this or that move is good or bad. GREAT learning.

Here's the really fun match of Mochy vs. Sander Lyloff  

• Day 1
  
• Day 2
• Day 3

And day 1 of the 2020 tournament (not that spectacular matches but much to learn from)

To hit or not to hit?

Recently I came across this position when playing Gnu Backgammon:

 

 

White to play 65

 

I thought that the big decision is: hit with 13/2* or play it safe with 13/8, 9/3. I hit on the 2, reasoning that I'm only slightly ahead in the race (4 points before the move), I have the stronger board, and gnubg has two blots in his home board, so if it manages to re-enter and hit me it should not be able to consolidate its board. I concluded that I'm well ahead if I hit.

On the other hand, if I play it safe I have no blot, but I risk that after 62, 55, 54, 53, 52, 44 gnubg gets out. That's 10 rolls that make the game even.

So I played the big blunder 13/2* (-0.21 equity compared to 13/8, 9/3, according to XG2's rollout).

Here's the dice distribution of gnubg rolls after my two moves:

 

Rolls after the best move, 13/8 9/3. Only 55 is dangerous (bringing gnubg's blot to safety)

 

 

Rolls after my move, 13/2. If gnubg hits me with a 2, XG considers me no longer a favorite.

Seems I overestimated the danger of gnubg escaping when I play safe - only a 55 diminishes my significant lead. I also underestimated my winning chances. I was expecting a small lead like 0.2, not the 0.8 XG assesses. 

My hitting move makes gnubg a favorite if it hits. That's also something I didn't expect. But if you actually look at the positions after every move 2+x, it shows that gnubg can always cover one of the blots and make a three-point board itself, giving us an almost symmetrical position (and costing me 23 pips in the race, obviously).

The basic lesson I learned: with a big roll like 65, don't consider the pipcount before but after the move. I was not 4 pips ahead but 15, which is quite a lead and which suggested a conservative play.

Why should I not drop here?

In my last match against XG2, I thought I was clever and XG2 refused to believe that I am. I'm still stunned why my thinking is wrong.

Match to 7, I'm leading 6:4, Crawford game. I roll an initial 62 which I consider poor. XG2 actually agrees, rates my position after that 62 as -0.3. 

XG2 does the automatic double, and I think "well I have three attempts to win that one point, I'm sure I get a better first roll than 62 in one of these two. I drop.

XG2 then tells me I've wasted 0.7 equity and played this like a total beginner. XG2 can be cruel in its assessment of us humans.

The match continued with a next game that I correctly dropped after two unlucky rolls, and a final game in which my first roll was a beautiful 42, so I could start with 8/4 6/4 which is a much better start than the above...

Update: it turns out (thanks to David Startin on Twitter) that my reasoning was, well, flawed. Yes, I only need to win one more game, but a game is more than a good start. So by passing I gave away a match where XG needed to win two games (the first to 6-6, then a final one), the first of which had a mediocre start for me, for a match where I got a better start but XG needed to win only one game, which clearly is a big blunder.

Backgammon HD by Wildcard Classics

Wildcard Classics confronts the user with 3 different Backgammons:

• Backgammon - classic board game - free, 2 in-app purchases (for 2.50€, a Christmas theme and something called Acey Deucey???). Contains ads.
• Backgammon HD which I‘m reviewing here - costs 5.50€, no ads.
• Backgammon Premium, costs 3.50€, offers the in-app purchases as well.

On top, if you use Backgammon HD you can‘t play online against somebody who uses Backgammon - classic board game.

Once you‘re past this confusion (which their website doesn‘t explain, primarily because the URL they provide in the App Store doesn‘t exist), Backgammon HD is a fairly nice game. Not in the same league as the best ones, but good in every respect for casual players.

Backgammon HD offers a beautiful UI with several different skins

Like many Backgammons, BGHD features three difficulty levels - easy/medium/hard. Let‘s have a look at „hard“. It turns out that the app plays a decent game of backgammon. It knows the value of builders, it tries to make points, it hits blots when it should. Of the 11 games I played against it it won one. (I‘m not an expert, btw). 

For confirmation I let XG (PC) play a game against BGHD. BGHD made a good number of correct moves, and a good number of blunders. Not the kind of hyper-blunders that lose a whole game like some other apps I‘ve reviewed lately, but blunders. Not unexpectedly, XG‘s assessment of BGHD was „Beginner“ with a PR of around 30.

If you‘re a casual player and if you look for a good casual player to play fun and challenging games against, this might be your app. If you‘re a Backgammon enthusiast who wants to get better, to learn from your mistakes, it‘s not.

BGHD offers the full backgammon incl. doubling cube. However it feels like the doubling cube was added as an afterthought. Doubling well is an art of its own, and BGHD‘s rule „if you‘re well ahead in the race, double or take“ fails so often so miserably. I‘m winning most of my games at 64 points.

The screenshot above shows the situation after the app re-doubled from 16 to 32. Yes, my pip count is 118 and the app‘s pip count is 104, so it‘s ahead in the race by 12 pips. But I have trapped a stone behind a 5 prime, threatening to make a 6 prime, and it has nothing! This is the time to resign, not to re-double. The game ended with me winning 64 points.

My greatest misses, part 1: a subtle detail on bearing off

Okay, I'm only an advanced level backgammon player dreaming of playing on expert level some day, but maybe my worst blunders can be instructive to other non-pros. 

Here's one you might enjoy. It might be trivial for you, and I'm embarassed that I missed it. I learned - situations that look like "there's only one move so don't even start thinking" are still worth thinking.

 

 

Here I didn't think. It was clear, there is only one checker that can move the 6, so I moved 8-2. Then I needed to move the 4, so I moved 3-off. XG tells me this is a horrible blunder worth half a point (0.44 to be precise). 

The right move is to move the 4 first - 8-4, and then bear off the 6 with 4-off, leaving no blot.

True Backgammon HD - BGBlitz on iOS

Not as old as the real pioneers of computer backgammon, but old enough, there since 2002, and still being developed and performing on top level: the strong backgammon program BGBlitz. For iOS, there are two apps utilizing the BGBlitz engine, and while I wasn't entirely convinced of Backgammon Gold, the second app, True Backgammon HD, shines on iOS, and is a worthy contender for the Backgammon crown on Apple's platform.

 

 

True Backgammon's leather board (iPhone)

  

Visually, and acoustically, TBHD is my most favorite Backgammon for iOS. The leather board above costs €1 in-app, but that's worth it. By default there's a nice wooden board and a just as nice metal board for you to choose. And while the clicking and dice-rolling sounds of nearly every other backgammon sound cheap and cheesy, the ones of TBHD are so enjoyable that I'm actually playing most games with sound on.

 

 

The free metal board. Notice how the triangle shape of the pips shines through the semi-transparent checkers? Nice!

The drawback of this clean, simple, beautiful design is that actually playing is a bit annoying. Want to know who leads in the race? Tap on menu to have the board scroll sidewards and reveal this info. Want to try out different moves (which is absolutely normal in Backgammon, even on professional level)? Tap the menu, wait for the board to scroll to the side, tap undo (you can also just move the pieces back to their previous location, like on a normal board, though).  

Another minor annoyance is the way the game supposedly beautifully scrolls up a bit when offering a double:

 

 

The wooden board, I'm being offered a double

Doubling is a critical, complicated part of Backgammon that requires full concentration. Frankly, it's beyond me why I can't see the full, normal board but only see the top half of the upper row. This might be considered a minor issue, but it's annoying as hell (particularly as my cube handling is vastly inferior to my dice handling, so I really need to concentrate there).

Moving the checkers on a tablet is a wonderful matter. You grab a checker and move it to its destination with a well-done drag-and-drop implementation. You can even drag-and-drop back and forth when trying out different moves, like on a real board) and multi-drag several checkers. By now, TGBG supports different ways to move, on devices of all sizes. I like drag&drop most because it feels nearly like a real board.

Yes, these are no major issues, but TBHD costs somewhere between €4 and €15 depending on your in-app spending habits. You need at least the basic game (€3.50) and the BGBlitz AI features (€6). At this price of admission I expect a rather flawless app.

 

 

For a Tron-ish Backgammon experience download a neon board (in-app purchase €1). Here I'm being offered a hint by the tutor.

 

If you purchase the BGBlitz tutor for about €3 (in-app), and you should, because an AI that destroys you without telling you where you're failing is a frustrating experience, then TGHD will give you the usual insights into your errors and blunders. This is nicely done, shows everything you get out of the other top-notch mobile AIs.

This is what it looks like: it tells you the preferred move + all the equities and probabilities that backgammon fans need to understand the evaluation. Also the game shows you visually on the board what the best move would have been. Same for doubling. And with BGBlitz being fairly strong, these hints can help a lot to tune your mental pattern matching abilities.

As we're at it: how strong is the app, exactly? I analyzed quite a few games with XG2 on the PC (roller++); result is that it plays about as good as the best humans on their very good days. It performs at a PR rating of about 1.5 to 2.5. Similar to BG NJ, slightly inferior to XG Mobile's super-human strength of about 0.5 PR. But honestly: you'll never notice. BGBlitz in this app makes no errors that are worth mentioning, and it will teach you to avoid your errors just as well as XG Mobile.

Just to be sure you take this with the right grain of salt: A low PR means better. World-class players manage to get a PR below 4. The currently strongest human player has a PR of 2.4. TBHD has a PR of somewhere around 1.8. It plays significantly stronger than any human can play.

It's a difference if you don't invest the money for the BGBlitz in-app purchase. XG rates me as an "advanced" player with a PR of about 11, and I'm soundly beating TBHD's free built-in AI on its hardest level. For newbies the AI is sufficient but don't expect to learn anything, improve your backgammon, or maybe get your mind blown by a state-of-the art AI destroying you over and over and showing you that luck plays a much lower role in Backgammon than you thought.

 

 

The (also) beautiful oriental board, scrolled to the left to make place for the menu.

For the sake of completeness: the app offers two-player "share a device" gaming and network gaming. I havent' tried out both. The app comes with pretty much all rules, but I haven't found Jacoby/Beaver. You can play money games or tournaments of every normal tournament length.

Double-clicking on the match score will open a "export and mail match as sgf". However often XG2 and/or gnu bg can't import the sgf due to some errors in the file. Hoping for a fix.

My summary: this is a great backgammon. With some small tweaks (maybe an "undo" / "show pip" gesture; single-click move on iPhone, different way to show "double?" dialog) it would be the best of the best. As it is, it's the backgammon of choice if you want the most beautiful top-notch backgammon on an iPad. If you want to play not only a world champ but a superhuman AI, go for XG mobile. If you like post-mortem analysis on the device, go for BGNJ.

Update: in the last months, TBHD has received a couple of great new features that catch up with XG Mobile. I particularly like the new game / match statistics (German here, unfortunately) that show you precisely how well you were doing in the current game and match, and whether the dice gods were looking down on you approvingly or punishing you with bad luck :-)

As a result, I‘m playing nearly all my “bot matches” with TBHD; the beautiful board, the nice drag and drop, the strong engine, and the insights in a single package are better than the competition.

Disclaimer: I’m not affiliated with the developer of TBHD, but after my first review he contacted me, took my suggestions seriously, implemented them, and I acted as a tester for them. So I might be more enthusiastic about the app than if I were completely neutral. I consider this just an additional feature: the developer cares and listens. (I’ve got different reactions to feedback I gave to other apps…) 

Playing online on Backgammon Galaxy

Let‘s start with a few confessions: I hate „pay to win“ games. I‘m not talking Backgammon here, but normal games you download on the App Store or Google Play. These games don‘t feel like they are designed for you to have fun. They are designed, primarily to milk you for virtual coins so that you buy virtual coins with real coins. I also hate annoying sales people, but seriously, who doesn‘t.

That said, I‘ve come across three kinds of online backgammon sites: 

1. Those that seem to have been added to an app as an afterthought (Backgammon NJ). They just don‘t feel fun to play.
2. Those that stem from the ancient days of online backgammon - FIBS clients, Backgammon Studio Heroes. They are free, feature-rich, fun to play but the user experience shouts „I was designed in the last millennium!“. And particularly with respect to FIBS: downloading a PC/Mac client? In 2020? Seriously?
3. Those that (and that brings us back to my confessions) want to make money when you play backgammon. Sites like Backgammon - Lord of the Board that SCREAM at you! Win! Click! Pay!

But now here‘s 4. Backgammon Galaxy, a web site for serious Backgammon fans that is beautifully designed, innovative as hell, plays as great on a PC than on a phone. It has a few issues, but wow I‘ve seen the future of online backgammon.

This is how it looks. @Ychan if you read this and want your name removed, just comment here.

Playing a game on Backgammon Galaxy (iPad Pro)

Backgammon Galaxy features a very distinct, very simple, IMHO very beautiful look and feel, with a futuristic theme and a blue-ish design. Rolling, moving, doubling is all done as you would expect; I never ran into an issue trying to move a checker to a place or doubling. Currently, only match-play is supported with matches running from 1 to 25 points. You can also select three different time controls - fast is 8 sec per move + 20 sec per point, normal is 10 sec per move + 1 min per point, casual is 15 sec per move + 3 min per point. I love casual, gives you plenty of time to think. But if you want time pressure, you can get it :)

Matchmaking is simple: one player creates a game by filling out a simple form, entering desired match length and clock. The game offer is shown in a list. Another player accepts the offer and the game starts.

The overview screen. Here you can join a match offered by somebody else or create a game offer yourself.

There‘s also tournaments, but I haven‘t tried them yet. If I understand there's a registration period and a start time and then you do knockout rounds.

 

 

A finished tournament

Unfortunately there are no private games. If you want to play against a friend, this isn‘t easy here as your game offer will be there for everyone to accept and usually it takes seconds only until you get your match.

Once you‘ve started a match you play a normal match of backgammon. I love the beautiful UI, I like how they implemented checker movement - tap on a checker to move the first die, then same for the second, tap on the dice to swap, if you prefer, drag and drop checkers. I‘ve got two minor complaints, though. First, I don‘t see an animation or similar that shows me where my opponent has moved; the simple UI also doesn‘t show me the last move. Sometimes I have to guess what exactly my opponent did. Second: I don‘t get the sounds. Sometimes the game makes tiny clicking noises for unknown reasons; the one sound it should make („opponent has finished moving, your turn“) it doesn‘t. So I sometimes didn‘t notice that it‘s my turn again. But that‘s minor really. Playing a game of Backgammon is extremely enjoyable on Backgammon Galaxy.

Once you‘re finished, you‘re taken to the total magic wonderful game-changing killer-feature of Backgammon Galaxy, the feature while I pretty much stopped playing anywhere else. The Match Result incl. rating adjustment.

Game result of my most recent game.

Have a look at this screen. You see not one but two crowns. One for the player who won the match. One for the player who played the stronger moves. Now the thing is, you need both crowns to have your rating improved and vice versa. Win by luck although you‘ve played weakly - no point for you. Lose a match where you played like a God (like my opponent in the screenshot) because your opponent was lucky - you won‘t lose a single point. 

There are discussions going on whether this approach is the right one; after all, sometimes the right move is the wrong move. For instance, if you‘re playing somebody who is much weaker you might want to play an aggressive move instead of the optimal move. Personally, not encountering the frustration of losing against a clearly weaker but luckier opponent and losing rating because of this is a major step forward in enjoying online Backgammon.

How do they achieve this? Behind the scenes, the supposedly strongest Backgammon engine, Extreme Gammon, keeps track of all moves, and determines how far you and your opponents have strayed from the best move. And after the match it will tell you.

Which brings us to another fine feature of Backgammon Galaxy: having your games reviewed by a strong AI.

At any time, you can bring up a list of all matches you've played. In this list you can either export the match to have it analyzed by XG or gnubg on your computer, or you can use the beautiful and pretty decent built-in analysis that you can see above. 

The UI shows the list of moves of each game in the match, along with differently colored dots that show where you or your opponent made a mistake. Tap on a move to see a detailed analysis of that move:

You need to learn a bit about equity (no big deal), but then this analysis shows you nicely what the best move would have been and how far off your move was.

Of course, this being a free online backgammon, you won't get the full power of XG to analyze your matches. For lowly <1750 ranked people like me, it uses XG with a 2-ply analysis. That is way enough and tremendously helpful. Still: if you download the match to analyze with a PC you get a much stronger analysis (XG spends 250 times more time analyzing each move).

A big construction site: in several places you come across pieces and bits that are under construction.

 

You can't play private games, you can't play money games, you can follow somebody, but that doesn't seem to mean anything (like "challenge a friend", getting updates about their progress etc), you can't create tournaments on your own, you can't use whatever the blunder DB is going to be etc. There's even a premium membership that, well, isn't there.

But that's not bad. I hope all those features will appear one by one. Until then what's there is well enough to play some wonderful games of backammon.

Backgammon Elite is not elite

 To start with the positives: Backgammon Elite has a really nice in-app concept - the basic game is free; you pay for additional Backgammon variants. No ads! And the board graphics are just as beautiful as the checker animations. Good job there.

My first game against Backgammon Elite. Nice graphics. But the engine... I‘m black. Notice how it didn‘t even seem to try to make a board?

If only it would play Backgammon. And if it would play the game, which it thinks Backgammon, reasonably well. But it doesn‘t. And it doesn‘t.

First: as said several times: the doubling cube is an essential ingredient of Backgammon. Without it you play the much less intricate game that Backgammon used to be before the cube was invented.

Second: Sorry to say this, but this game plays like a beginner. And there are many apps out there who don‘t. You cannot learn anything by playing against this app. You can learn so much, including how enjoyable the game can be, from eg. XG Mobile or Backgammon NJ.

My second game against Backgammon Elite. Still nice graphics. I‘m black. Notice how, again, it doesn‘t attempt to make points?

Many years ago when I was studying computer science, I actually spent a week or so writing a backgammon engine. It was playing poorly. I was beating it soundly most of the time. But it played much better than this engine. You know, basics like „Low stacks have a higher value than high stacks“ or „ownership of the 5,4,3,... point has some value or „the value of a blot is how many good points it can help slot - the number and hit probability of opponent stones“ are real simple to do. So it‘s a total mystery to me how the developers of Backgammon apps invest so much time into a beautiful presentation of such a poor engine.

You‘re still here? Then a few more facts about it. It doesn‘t feature online gaming, it doesn‘t feature matches, it has basic „who has won how often“ statistics, and a „2 players pass the device back and forth“ two player mode.

Off to the next one.

Big Brother is rating you - the next generation of online backgammons

Okay, maybe I'm the last person on earth to learn about this. Apologies if I'm selling old news as new news to you then. For me, spending some hours with Backgammon Galaxy (henceforth Galaxy) and with Backgammon Backgammon Studio Heroes (henceforth Heroes) was an eye opener on where online Backgammon is heading.

The two sites are very different. I will review them individually later. For now let's say that Galaxy is fresh, clean, beautiful, but also under construction and lacking many features, while Heroes is bursting with features, but under a presentation layer that looks like Windows 95. Also, the two use the "Big Brother", but very differently.

The "Big Brother" is Extreme Gammon (XG), the PC program that I'm also using here in order to tell you objectively how good Backgammon apps are playing. Like Gnu Backgammon, but much speedier, XG plays a God-like backgammon far above what the best humans can pull off.

Now suppose you can integrate an omniscient Backgammon God like XG into your online site. What would you do?

Backgammon Galaxy

Galaxy goes the non-compromise, clean, beautiful way. Honestly, I sat there in shock and awe when I saw what it does. Unlike every other site (Chess, Go, Backgammon) that I know, it doesn't adjust your online rank upwards when you win a match or downward when you lose a match.

Your rank improves if you

a) win a match

b) have objectively played stronger than your opponent in this match.

Your rank declines in the opposite case - you lose and you have played objectively weaker.

Bummer. 

What this does it it takes the frustration (loser) and embarassed feeling (winner) out of the game, when you win a match against somebody who was outplaying you, but just didn't have luck on her side this time. Because if you win, but performed worse than your opponent, nothing changes.

After a match you get to see a screen that tells you how you fared (e.g. won a 5p match 5:2), and how you performed (e.g. you had an intermediate performance of PR 11, your opponent an experienced performance of PR 9). And then it will tell you how it updated your rating. In our example, nothing will happen. You've beaten your opponent, but she played stronger than you, so the Gods of Backgammon decide you get no reward. Here is their own explanation of their system.

 

Here‘s an example of a „1 point game“ I recently lost. My opponent got the one crown for winning the game. I got the other crown for having played the better moves. As a result, the Galaxy rating of both of us stayed unchanged.

A note on PR: this "performance rating" is an invention of XG that by now is even used to rate and rank the performance of the world's strongest players in BMAB , the Backgammon Masters Awarding Body. The PR reflects the number of errors you objectively make. A "super grandmaster" has a PR of less than 2.5, an advanced player between 6.5 and 10, a newbie 20+. This is, of course, varying largely in different matches. Some matches only give you easy to assess positions, some are so tricky that you blunder over and over (worst: you continuously misjudge whether to double and the situation stays similar for several moves). I've had single games where I was ranked "world champion", I've had single games where I was ranked "distracted". Even Mochy, arguably the strongest human at the time I'm writing this sometimes plays games on "expert" level only.

Of course you can critizise this approach -  in some cases you play in a certain, non-optimal way on purpose, maybe you play against a much stronger player so you play more conservatively, or maybe you're behind and feel the need to play extra aggressive. But in my opinion the advantages, the good feelings you get by removing the "DAMN I WAS SO UNLUCKY" results from the equation, outweigh these massively. And you know that you have to find the objectively best moves to improve your ratings.

Backgammon Studio Heroes

(man this is a clumsy name, I still always have to look it up everytime I write it. Why not just call it "Backgammon Heroes"?)

Heroes goes a more compromising way. It offers you totally normal backgammon matches, albeit with the Gods of Backgammon (XG again) informing you about your skill and luck, which is already good - sometimes you feel that you've been much better but totally out-lucked, only to learn that she wasn't that lucky and you weren't that much better after all.

But Heroes has, among its myriad of match types, a few that are truly special.

 

 

Heroes' match types (in the overall screen; see what I meant about Windows 95?)

 

The coolest one is "PR Match". In this match you play a normal match of backgammon. But the winner is not who won the most points. The winner is who had the better PR in playing the match.

Cool is also "Blundergammon". XG considers a move that is 0.08 points of equity or more worse than the optimal move as a blunder. In this match type, the player who makes the first blunder immediately loses.

So while I personally prefer Galaxy's no-compromise, beautiful, clean approach to ranking, Heroes is also an interesting approach, giving you the traditional backgammon experience (if you want, even without the big XG brother watching you - that's the "plain match"), and lets you choose to play for PR or non-blunders instead of points.

But both sites show how much you can do if you have a god-like engine running in the background of every single match. Do I need to mention that both sites also give you strong XG-based analysis capabilities that help you improve your game. 

Highly recommended. (warning: on an iPhone, Heroes looks like shit, you need a tablet or computer to enjoy the game; Galaxy looks very beautiful on any device, but this I'll cover in reviews later)

 

Updated my reviews of a few apps

Nothing new today, but quite an overhaul of my reviews of XG Mobile and Backgammon NJ, which still rule planet iOS Backgammon. Don't let them tell you otherwise. 

After playing a few more matches with XG Mobile and having the PC XG analyze it I am amazed how strong XG Mobile is. Far better than every human being.

I also played some matches against Backgammon NJ. I love how you move checkers there. Strength-wise it's not on par with XG Mobile, but on par (in fact a tiny bit better) than the currently highest rated human. Good enough for me, I guess.

Also added screenshots with different visual designs to the two apps. Enjoy!

 

 

XG Mobile can look quite different to its standard design, if you want.

Get my men out of here, sergeant! (Backgammon Now Review)

Backgammon Now was featured (maybe just for me) on the Apple Store and got some fairly good comments there so I had to try it out.

The app has good graphics, a reasonable price for getting rid of annoying ads, and implements pretty much the full rules of Backgammon. I couldn't check whether they feature the Crawford rule, though. Because I win all my games with something like 32 points. And this brings us to the heart of Backgammon Now's problem: it has no clue about Backgammon. 

Have a look at this:

This is the position after three moves - on hard difficulty, which is the strongest that the app offers: black (Backgammon Now) rolled 65 and actually pulled off the right move (that happens once in a long while), moving 1/12. I rolled 32 and moved 13/11, 13/10. And then the app rolled a beautiful 22. It could have gone 19/21(2), 12/14(2), or maybe 1/5, 19/21(2). But of all available moves it picked 1-9, taking its back man and putting it into three direct hits.

And this is the general pattern of how BG Now plays: it wants to get the back men out. Now. It has no understanding whatsoever about slotting, making points, direct vs indirect hits. It just piles up its men, and when it can hit, it hit's. At least that's something it can do.

Here's a few more beautiful positions it got itself into.

 

Did I mention it doesn't know when to double as well? It seems its doubling rule is "if my pip count is smaller than his pip count I'll double or take". It kept re-doubling although it was trapped behind a 6-prime. 

 

Backgammon now handing an initial roll of 62. Moving the 1 checker doesn't achieve anything. A builder on 14 would have been great, or slotting 21. But no. I HAVE TO GET MY MEN OUT!

This is a typical position that we always end up with. I'm building a good base, while it just ignores its home and outer board completely. It then gets hit over and over and has half of its checkers behind.

 

I try to be fair when reviewing apps. I'm a computer scientist myself. I know the amount of labor and passion it takes to create an app. But either the developers themselves have no clue about Backgammon (then they shouldn't develop a backgammon or learn how the game works first), or they don't care about releasing an app with the worst AI on the App Store. Then they should go back to the drawing board.

If you're a total beginner then ... no, not even a total beginner should use this app, because it won't teach him anything but how not to play Backgammon.

Why is this the best move? (another one)

 Okay, dear readers, I'm sick at home with an awful cold (plain normal cold, nothing covidish) and have time to play and analyze a few Backgammon games. Here's another position that stunned me:

I'm white and have to play 61. The move 8/2, 3/2 looked straightforward to me: it extends my board to 5, it doesn't endanger my back men - the men on 19-17 seem to threaten building a bad board, fast, if they get a blot to hit, and it only leaves a single blot 1 away from 7.

XG doesn't like this move at all. In fact it believes that the best move keeps the game equal, while mine brings it down to -0.28, so I lose a third of a (1 point) game with this move.

And the move it suggests is... 22/16, 3/2. WHAT? 

Let's look at the dice distribution for my opponent after the two rolls:

 The first image shows my winning chances for each roll that the opponent can do for the XG move, the second one does the same for my move. Example: first image, column/row 3/4 shows +0.082. This means that after XG's move if the opponent rolls 34, the outcome will be a 0.082 in my favor. 

A few interesting observations: other than I thought, only 33, 44, 55, 66 are rolls I have to fear when playing the XG move. Moves like 43 that I was fearing are probably not so bad, because black will only have a 3-point board and a blot that I might hit as well. 31, 41, 54 that I also feared (hitting me and moving on to cover its own blot are still not too bad for me for the same reason - a 3-point board isn't too bad.

On the other side,  the same doubles 33, 44, 55, 66 are similarly bad in my move. After all, my two men are trapped behind black's little prime; if black moves his trailing men safely forward with the double, he can wait for my men to pass the prime.

My lesson for the day: if confronted with an intimidating prime, look whether things are really as bad as you think.

 

Why is this the best move?

 Have a look at this simple position:

Without much thinking, I played 24/23, 13/11 here. This adds another builder, splits my back men, looks like the straightforward choice.

However, XG stubbornly recommends 24/21 as the best move. Mine is fine, too (-0.05 worse). But still, why should I not put a builder on 11?

The answer is probably that the builder doesn't help much - I already have the 5 point, 11 is too far away from the four point, and as my 8 point is stripped and my opponent has a stone on 2, building the 7 point in the next move with e.g. 11/7, 8/7 is not an option. Also I underestimated that black has some really great rolls like 55 or 44. XG (on Roller++) also says that 13/10 would be slightly better than my move, because then the builder on 10 actually has something to build - the 4 point.

On the other side, black has made the 18 point, threatens to make the 20 point, so helping one checker to escape is reasonable.

Sometimes the straightforward move is not straightforward.

Do you agree? Any thoughts are appreciated.

Is Backgammon trivial compared to Chess and Go?

 Some time ago, in a social media service that doesn't exist anymore, I made a bold claim. I claimed that Backgammon is as complex as Chess and Go. People didn't agree, and without evidence I found it difficult to make my point. 

Let me try this again, here in this blog :-). In Backgammon you come across several kinds of complexity, like making the right strategic decisions.

Making the right strategic decisions

Let's look at a few tricky positions in my recent (lost) match to 13 against Extreme Gammon Mobile (print and analysis by XG for the PC, Roller+) and see how tricky things can get in Backgammon.

 

 

Here I'm white and rolled 63. I have a couple of issues: there's the blot on 21 that I want to save before B hits it and occupies more slots in its home board. I've got no good home board whatsoever, and an awful stack on 6 that I need to shrink. So I thought it would be logical to play 21/15, 6/3, getting my blot in a safer place where XG can only hit it with a 3, and go for the 3 point with limited danger (only a 2 can hit me). Okay, in most cases it will hit me, but I hope to achieve one of my goals: save the blot or make an additional point.

I was wrong, but not by much (0.1 points of equity). The best play is more conservative: 21/15, 11/8. Probably my mistake was that if I get hit on the 3 point, both parts of my plan are moot - I have not made a point in my home board, and I have another blot left behind that it can attack.

 Same game, one move later - XG rolled 63, hit my blot with 12/15, and moved 4/10. I rolled 41 and found myself in this tricky position:

Black has a lot of blots that I might want to hit. 20 with Bar/20*? Looks solid, lets me attack the blot on 15 in the next move, but weakens my blot on 3 even more (any 3, any 2 will hit it). Or hit on 10 with Bar/21, 11/10? I have nice builders on 11 and 10 then, my blot on 21 can hit its blots on 20 and 15. Or maybe I play conservative Bar/24, 7/3, have a 2-points board, save my blot on 3, and look which of its many blots I can hit on the next move?

I decided for Bar-20*, and yay! I was right. Actually my move and the conservative Bar /24, 7/3 get nearly the same score by XG, although they will lead to very different games. My other candidate move, Bar/21, 11/10 gets a beating by XG - it loses nearly 1/3 of the game (-0,29).

Calculating risks properly

A part of Backgammon where I really suck is calculating risks properly. Take a look at this position.

That's a fairly easy one, you would guess. And I still managed a major f..k up. My reasoning was: yes, this is my chance to fill my 1 spot, make a 5 board by playing 8/1, and all for the minimal risk that he might hit me on 8. XG tells me I lost half a point with this decision. The best move is the simple and conservative 13/8, 13/11. Why is my move so bad? You have to look at the dice distribution for black's next roll to understand:

 

 The first image shows how the game will evolve after the best move. The second image shows my move. Two things can be seen: first: I underestimated that clearing the 13 point and keeping the 8 point is really important. All my followup moves are some 0.3 worse than necessary. And if it hits me with 11 (which it can also use to make the 21 point, I've moved my chances from 0.6 to -0.7. Keeping track of such "anti-jokers" is really important, because they can turn around the game in a single move.

That doubling cube

The doubling cube is a fiendish addition to Backgammon. I'm losing much more points by messing up my doubling than I do with poor moves. There's a couple of reasons: I forget to double because the action is so intense. I take where I should pass because I underestimated "anti jokers" like the one above. And I pass where I should take because I'm worn out and just want to stard a fresh game. Let's see a few examples.

 For me, the situation was clear. There is a good chance that I can't enter my stone against XG's 4-point board. And if I do and hit at 23, I have this blot on 7, need to get 3 stones back. Pipcount is nearly equal. A clear "no double". 

XG says I should double. I'm still confused why, when writing this. Probably I didn't consider that this is a match to 13, and in this position I'm trailing 4-10. If I double and can't enter, re-doubling is not as grave as normal because 4-14 is the same as 4-13. And as the AI believes I'm slightly leading, and have a few killer moves (2 + anything, 11), I should double.

Let's see another one:

 

 Here I didn't double. I'm well ahead, but I'm scared of moves like 21 that would give XG two shots at my poor blot, while having a 5 point board. XG tells me the correct cube action actually is "double/pass".

It's roll distribution looks like this:

 Yes, there are a number of really awful rolls for me. I didn't even think of the horrible 6-6 where I can't move at all, or 6-4, 4-4 where my blot not only stays in danger, but actually has to stay at its current spot, 4 and 6 points away from black stones, very likely to be hit. But most moves bring me to certain victory, and if your winning chances outrank your losing chances like that, you should double.

Stamina

Now this has nothing to do with the game itself, but I notice that in a match to 11 or 15 my performance degrades as we progress. In Backgammon, about half of the positions you get are totally straightforward. Some are immediately visible as tricky, and some look straightforward but you manage to not see an important move. And I, personally, manage to often miss important moves less later in the game.

 

Tactics - not so much

In my opinion, Backgammon delivers the same challenge concerning strategy, planning and mental stamina as Go and Chess do. But other than chess and go, Backgammon offers only very limited tactics. In Chess you need to play a number of likely move sequences in your mind to see how the game will evolve with this or that move. In Go, this is even more prominent, "reading" a position well is key to win local battles. In Backgammon, there's the dice, and calculating the outcome of the game with every single die combination is not something you can pull off. Really good players can detect critical "anti jokers", and the best bots look ahead 3 plys.

My conclusion

Backgammon is a much faster game than Go or Chess. Tournament chess and go matches use something like 3 minutes per move. Games typically last 6 hours. In Backammon, clocks are set to 2 minutes per match point. A tournament match to, say, 11 points then lasts for max 44 minutes. A single game lasts for something like 3 to 4 minutes. After all, the lengthy calculating of variations is not a part of backgamman, and this is the part of Chess and Go that eats most time.

And that's my conclusion: In terms of strategic complexity, Backgammon is a worthy peer to Chess and Go. But the combination calculations / tactics in those games are just not there in Backgammon...