24 June 2018

The Sociology/Psychology/Philosophy of Chess

Next up in the series on The Sociology of Chess (November 2016; last seen a month ago in Sunday Series) is a video titled 'The Psychology of Chess'. One of the comments says,

Interesting insights. I feel it should called 'The Philosophy of Chess' though.

Sociology, Psychology, Philosophy -- take your pick.


The Psychology of Chess (14:16) • 'Published on May 22, 2018'

The description starts,

This video explores the psychology of chess, especially with respect to how chess can offer us valuable life-lessons.

At 1:00 into the clip, a slide says, 'A Road-map of this video':-

1:11 - Finding a worthy opponent
2:38 - Learning to win and lose gracefully
4:25 - The trap of seeking ego-gratification through winning
5:04 - Chess as a spectator-sport
6:13 - Chess as an aesthetic event
6:54 - Discovering one's personal style
8:24 - The sting of defeat and its lessons
9:55 - Hubris Lesson #1: Underestimating your opponent
11:15 - Hubris Lesson #2: Assuming you already understand the position
12:03 - Hubris Lesson #3: Learning to see the big picture
12:47 - Final Lesson: "Love while the night still hides the withering dawn"

At 1:05, another slide says, 'Thinkers, authors and chess-players cited in this video':-

Mikhail Tal (0:22)
Erik Erikson (1:38)
Carlos Castaneda (1:50)
Rudyard Kipling (3:38)
Siegbert Tarrasch (6:30)
Josh Niesse (7:38)
Savielly Tartakower (10:23)
Emmanuel [Emanuel] Lasker (11:30)
Nightwish (13:36)

For more about the speaker in the video, see Eric Dodson, Ph.D., Department of Psychology, College of Social Sciences, University of West Georgia.

22 June 2018

Analyze Your Own Images

In last week's post, Search Your Own Images, I noted 'First stop: images.google.com', fed it a few images, and finished with a test on a special image:-

After I uploaded the image, Google declared, 'Best guess for this image: donald trump chess', where the first link (from 'About 25.270.000.000 results'!) was 'Does Donald Trump play chess?' (quora.com).

Second stop: How Image Recognition Works (upwork.com). After a brief, concise introduction to the technology behind image recognition, the article mentioned,

A number of APIs have recently been developed that aim to allow organizations to glean insights from images without requiring in-house computer vision or machine learning expertise.

This was followed by a list of three specific services:-

  • Google Cloud Vision
  • IBM Watson Visual Recognition
  • Clarif.ai [NB: clarifai.com?]

That first suggestion led me to Vision API - Image Content Analysis (cloud.google.com/vision), which said,

Google Cloud Vision API enables developers to understand the content of an image by encapsulating powerful machine learning models in an easy to use REST API. It quickly classifies images into thousands of categories (e.g., "sailboat", "lion", "Eiffel Tower"), detects individual objects and faces within images, and finds and reads printed words contained within images.

I fed last week's Donald Trump 3D-chess image to 'Try the API' and it returned the following analysis.

The 'Faces' tab is shown in the image -- 'Joy: Very Likely', 'Confidence 94%'. The next tab, 'Labels', told me:-

96% Indoor Games And Sports
96% Chess
94% Games
94% Board Game
81% Tabletop Game
73% Recreation
71% Chessboard

The next tab, 'Web', included (1) 'Web Entities':-

1.0731 Donald Trump
1.0593 Chess
0.7329 United States
0.7046 Three-dimensional chess
0.6912 Brilliant Chess

And (2) 'Pages with Matched Images' with dozens of links, of which the first two were:-

The 'Document' tab drew a green rectangle around the lowest of the three chessboards, with no further explanation. I suppose this is related to the service that 'reads printed words contained within images'. The 'Properties' tab returned 'Dominant Colors' and 'Crop hints', while the 'Safe search' tab was the most curious:-

Adult: Unlikely
Spoof: Very Unlikely
Medical: Very Unlikely
Violence: Very Unlikely
Racy: Possible

I would say, 'Spoof: Very likely' and I have no idea where 'Racy: Possible' comes from. As for the last tab, 'JSON', this returned a text list to import the image's attributes into another process.

Who knew there was so much information in a single photo? Google knew. Google knows everything.

21 June 2018

Breaking the 2700 Barrier

No, I'm not talking about achieving a 2700 rating. I'm talking about post no.2700 on this blog. That distinction belongs to my previous post, Chess Lions in Leuven, which makes this current post no.2701. Before I achieve a 2700 rating, I first have to reach 2300, but since I stopped playing competitive crossboard chess many years ago, that's not going to happen anytime soon.

How many players in the world are rated 2700? I blew the dust off my collection of historical rating files and counted the players for 2018. While I was doing that, I also counted them on the first international rating list in 1971 and at the start of every decade since then (1980, 1990, etc.), where the lists were all published in January of the respective year. The following chart shows the counts I recorded. For good measure, I added a similar count of 2600 players (which includes >2700).


Left: >2700 players
Right: >2600 players

On the 2700 side, the only player breaking the barrier in 1971 was a certain 'Fischer, Robert James' (USA) at 2760. Second on the list was 'Spassky, Boris V' (URS) at 2690. The two 2700-players in 1980 were 'Karpov,An.' (USR) 2725, and 'Tal' (USR) 2705. In January 2018, there were four players rated 2800+, 40 rated 2700-2799, and 211 rated 2600-2699.

The increase in world class players through the decades has not been linear. There were more 2600 players in 1971 than in 1980, and the largest increase in both rating bands took place in the decade 2000-2010.

A few years ago, in Next Short Draw: 2750 (March 2013), I was critical of GM Wesley So for manipulating his rating to break through 2700 Elo. According to his Rating Progress Chart (ratings.fide.com), he subsequently broke through 2800 in January 2017. A 2800 FIDE rating is an amazing achievement for any player. To break 2800, all I have to do is write another 100 posts.

19 June 2018

Chess Lions in Leuven

The month of June means the end of spring, the beginning of summer, and the Grand Chess Tour in Belgium. According to my report from two years ago, Chess on Belgian News (June 2016), the recent rapid/blitz tournament was the third running of the Belgian leg of the event, formally dubbed 'Your Next Move'.


Photos: Massimo Marchetti (lafamily.eu)

The large posters in the right image were placed outside the elegant Leuven City Hall. The black poster reads,

The World Chess Hall of Fame (WCHOF) celebrates one of the world's oldest and best-loved games through vibrant, engaging exhibitions and creative programming. A not-for-profit, collecting institution situated in the heart of Saint Louis, Missouri, the WCHOF houses both the U.S. and World Chess Halls of Fame, which honor the accomplishments of the game's finest players and contributors.

The organization presents exhibitions that explore the history of the game, as well as its remarkable impact on arts and culture. These unique shows and programs are designed to appeal to both the chess novice and expert, defying expectations and enhancing knowledge of the game. Along with those of its sister organization, the Saint Louis Chess Club, the WCHOF's activities have distinguished Saint Louis as a national and international chess destination.

Mind. Art. Experience. 4652 Maryland Avenue, Saint Louis, Missouri, USA (314) 367-WCHF (9243) | worldchesshof.org

The white poster reads,

GRAND CHESS TOUR
ART of CHESS 2018

Grand Chess Tour: Art of Chess 2018 is the second edition of a traveling exhibition that fulfills the global mission of the World Chess Hall of Fame (WCHOF) by bringing artwork and artifacts to many of the stops of the Grand Chess Tour. This show includes highlights from the World Chess Hall of Fame (WCHOF) as well as loans from Dr. Jeanne and Rex Sinquefield, Purling London, and the organizers of the events in the Grand Chess Tour. The permanent collection of the WCHOF contains trophies, photographs, score sheets, periodicals, chess sets, and other artifacts related to significant players and events from chess history. The exhibition is inspired by the WCHOF's mission and its 2018 shows.

Among the highlights of this exhibition is an exquisite Hungarian chess set adorned with pearls, amethyst, and jade. Also on view is an enlarged version of the 2013 Sinquefield Cup chess set, created by Frank Camaratta as a gift for the founders of the Saint Louis Chess Club, Dr. Jeanne and Rex Sinquefield. These are supplemented by photography from last year's Grand Chess Tour and the trophies for several of the competitions, spotlighting the historic nature of these events.

Images from left to right: [top of the poster] Magnus Carlsen Winner of the 2017 Grand Chess Tour, 2017, Photo by Lennart Ootes; • Sinquefield Cup Imperial Chessmen, 2018, Photo by Michael DeFilippo; • Fabiano Caruana After Winning the 2018 Candidates Tournament, Photo by Nick Dunaevsky, March 28, 2018. • Collection of the World Chess Hall of Fame, (c) Nick Dunaevsky, official photographer of World Chess.

Wesley So won this year's Leuven YNM event. Details can be found on 2018 Your Next Move (grandchesstour.org).

18 June 2018

Where's Leela?

Where was I on the Leela series? My previous post, The Lineage of Leela (11 June 2018), was a filler. I spent my time on that post browsing the Leela forum, trying to understand more about the Leela technology. In the post before that, Understanding Leela (04 June 2018), I promised,

There are also many technical threads that discuss different configurations for the Leela client or that delve into the structure of the network(s). Let's look at those another time.'

The top of the main project page, LCZero, currently has an undated message that says,

Net has been replaced with a new bootstrap net with 50 move rule normalization and learnable batch normalization beta bias. It will likely be weaker to start but will hopefully grow stronger quickly.

What does that mean? Maybe I'd better retrace my steps, because it looks like I missed something. I'll start by looking at the links on LCZero's left sidebar (in order).

*** Getting Started: This goes to

Home · glinscott/leela-chess Wiki · GitHub
https://github.com/glinscott/leela-chess/wiki

That page says, 'A.O. edited this page 11 days ago · 26 revisions; Moved to https://github.com/LeelaChessZero/lc0/wiki'

Welcome to the Leela Chess Zero wiki!
https://github.com/LeelaChessZero/lc0/wiki

That page looks like the instructions I followed for Installing Leela (21 May 2018), so at least I'm on familiar ground.

*** Forum: This also looks familiar. The link goes to:-

LCZero - Google Groups
https://groups.google.com/forum/#!forum/lczero

The top of that page says, 'This forum is for all the discussion about LCZero - https://github.com/glinscott/leela-chess'. That github.com page says,

GitHub - glinscott/leela-chess: **MOVED TO https://github.com/LeelaChessZero/leela-chess ** A chess adaption of GCP's Leela Zero

Following that link leads to:-

GitHub - LeelaChessZero/lczero: A chess adaption of GCP's Leela Zero
https://github.com/LeelaChessZero/lczero

Note that the 'leela-chess' has redirected to 'lczero'. The main github.com page is:-

LCZero · GitHub
https://github.com/LeelaChessZero/

*** Github: This leads to the same page that I just looked at: github.com/LeelaChessZero/.

*** Networks: After more internal links -- 'Matches', 'Active Users', etc. -- there is another external link...

*** LCZero Chat: Here under #dev-log, I found the same message that is on LCZero's main page: 'Net has been replaced with a new bootstrap net [...]'. Signed, Chad - 10/06/2018.

Where's Leela? The main Github page has moved, but the rest remains as before. The chat looks to be a preliminary source of technical discussion to the forum.

17 June 2018

Impossible Backgrounds

Starting with last month's post, The Chess Waste Land, I moved the Featured Flickr Photo series from a fortnightly to a monthly schedule. This gave me plenty of images on my short list from which to choose. I chose the following.


The world of chess © Flickr user jaci XIII under Creative Commons.

The description for this image mentioned,

After: Rafal Olbinski. Chess photo is wallpaper.

Out of curiosity, I fed the image to the images.google.com service I discussed in Search Your Own Images. It told me, 'Best guess for this image: 3d chess'. One of the links for the image went to 4 Games Like 3D Chess for Linux (topbestalternatives.com), where there is a well known image of chess pieces on a sphere similar to the featured photo above.

In fact, further inspection of the two images convinced me that they were essentially the same. The other image is red instead of blue, but the same pieces are all on the same squares of the sphere. That must be the wallpaper mentioned in the description. Feeding the red image into images.google.com told me 'Best guess for this image: impossible backgrounds'. Can't argue with that.

15 June 2018

Search Your Own Images

In last week's post, An NN for Chess Images?, I used a photo from my archive and wondered,

Could a network recognize that this is not a photo? Or (without being told) that it shows dogs. My first step should probably be to make some sort of survey of what software and services are available.

First stop: images.google.com. This simple tool allows search on an image via a number of input paths. The most straightforward are:-

  • Copy and paste the URL for an image, and
  • Upload an image.

Both of these are accessed by clicking the camera icon in the search box on images.google.com. I fed it the link for the painting/photo used in 'NN for Chess Images', and it gave me three results. Two of these ('Best guess for this image') were pages about Dalmatian dogs (well done, Google!) and the third was a link to the home page for this blog, where the original post is still displayed because it is among the most recent.

The results also included a dozen thumbnails for 'Visually similar images'. None of the thumbnails showed a Dalmatian, but all of them had the same sort of pink background shown in my original photo. (Remember Chess in the Pink, April 2018?) Unfortunately, I can't give a link to the Google results, because it is based on some sort of encryption technique that probably includes details about its origin (i.e. me).

After that little experiment, I fed Google image search the link to the image used in 'Mystery Painting' on eBay (June 2017). This is a painting that pops up occasionally in different places, but where I have been unable to determine its origin. This time the Google results could find nothing more interesting about the image than that it showed a 'picture frame'. Under 'Visually similar images', the dozen thumbnails displayed different paintings in picture frames.

Given a similar image used in my original Mystery Painting (December 2007), Google determined that it showed 'edelen middeleeuwen', which another Google service translates from Dutch to 'nobles middle ages', which I translate to 'nobles from the Middle Age'. Along with a couple of links to Dutch-language pages, the thumbnails showed one more copy of my original image.

The Google image search clearly has limitations on its usefulness. As a final test I used an image which has been sitting in my 'Junk' directory for over a year, not because the image is junk, but because I didn't know what to do with it.

After I uploaded the image, Google declared, 'Best guess for this image: donald trump chess', where the first link (from 'About 25.270.000.000 results'!) was Does Donald Trump play chess? (quora.com). About half of the thumbnails were variations on my uploaded image; the other half were people playing chess, including two showing Garry Kasparov. The Trump image is obviously photoshopped, where a Star Trek 3D chess set has been pasted over whatever Trump was holding in his hands.

According to a database I constructed, my collection of images (mostly) from eBay has about 20.000 examples. Since uploading a small percentage of these one-by-one would be an exercise in tedium, I'll have to find a better solution.

14 June 2018

2018 USCF USChess Executive Board Election

Has it really been three years since I last posted about U.S. chess politics? So it seems: 2015 USCF Executive Board Election (July 2015). In 2016, there were two candidates for two positions, so I didn't bother voting. In 2017, there were eight candidates for four positions. I intended to vote, but something went wrong:-

Subject: US Chess Executive Board Ballot Mailing
Sent: Saturday, July 1, 2017

Dear US Chess Voting Member, Due to an error by our mailing agent, your Executive Board Election Ballot was not mailed to you at the scheduled time. Your ballot has now been mailed to you as of Friday, June 30 via overnight or expedited mail (depending on your location). The deadline for receipt of your marked ballot is Tuesday, July 18, by 3 p.m. CDT. Full instructions for marking and returning your ballot are included in the envelope being sent to you. Our apologies for any inconvenience.

Whatever went wrong the first time became a double blunder, because I never received the second ballot either. I was reminded of this when I received the following postcard earlier this week.

The card says,

Garry and Rex have teamed up to make a difference for American Chess by organizing and supporting premier chess events. Today they have issued this joint statement: "We are pleased with our partnership with US Chess. To keep the positive momentum for Chess, We recommend and endorse the executive board members and candidates with track records of positive achievements."

WE URGE YOU TO VOTE FOR:
Randy Bauer: FORMER US CHESS VICE PRESIDENT & VP OF FINANCE
Ryan Velez: US CHESS ORGANIZER & TOURNAMENT DIRECTOR

REX SINQUEFIELD, Chairman Chess Club & Scholastic Center of Saint Louis • Sponsor of 5 U.S. Championships; Sponsor of 5 U.S. Women's Championships; USA Olympiad Team Sponsor; Young Stars [smudge]; Sponsor Chess in the Schools, Chess in Education; Sponsor World Chess Hall of Fame; Sponsor U.S. Chess Hall of Fame

GARRY KASPAROV, Chairman Kasparov Chess Foundation • Former World Champion; USA Olympiad Team Sponsor; Women's World Team Sponsor; All-Girls National Sponsor; Young Stars -- Team USA; Sponsor Chess in the Schools, Chess in Education; Sponsor of Regional Scholastic Championships Greater NY, Greater Chicago, Greater Mid-Atlantic

This year there are four candidates for two positions. I don't follow U.S. chess politics closely enough to know why the two recommended candidates have better 'track records of positive achievements' than the other two candidates, but what's good enough for Rex Sinquefield and Garry Kasparov is good enough for me. With an American player competing for the World Championship in November, U.S. chess needs the strongest leadership possible.

Will I receive a ballot this time?

***

Later: Oops! Looks like I goofed. When I copied the title from the 2015 EB post to this post, I forgot that the U.S. chess federation had since changed its name from USCF to USChess. Strike one!

12 June 2018

USchess in Podcasts

Starting with the April 2018 issue, Chess Life (CL) has a new online feature, 'Cover Stories with Chess Life'. These are presented as a podcast interview with the author conducted by CL's editor, Daniel Lucas. The first talk was with FM Mike Klein of Chess.com, who wrote the cover story introduced on the left ('cover art by Paul Dickinson'). That first podcast, plus the two podcasts conducted for the May and June issues of CL (with Al Lawrence and GM Ian Rogers), can be found on Podcast Archives (uschess.org).

I'm a big fan of podcasts, and after The Week in Podcasts (February 2018), this is the second time this year I've posted about them. Maybe that's because I also like listening to the radio while I'm doing something else. Having said that, I have a couple of problems with the podcast format in general.

The first problem is that podcasts require concentration. I can't concentrate on a second task -- like writing a blog post -- while I'm listening to a podcast.

Driving also requires some concentration, as does manual work like painting a room, but listening to music at the same time takes no concentration. Listening to a talk show or a news program occasionally requires concentration ('Wait a moment. What did they just say?'), but my focus can shift rapidly because the primary task, like driving, doesn't always require full concentration.

I simply can't do two simultaneous tasks that both require near-full concentration. Why not just concentrate on the podcast? Because listening to a podcast is not like watching a video. Unlike a video, a podcast doesn't engage my eyes, which means that my visual attention is constantly wandering to something else. A transcript might be useful.

The second problem is that podcasts are difficult to quote, although here I have the same problem with videos. If I want to introduce a short quote from an audio source into a blog post, I have to listen to the audio and write what I think I'm hearing, then iterate the process several times to make sure the quote is accurate.

What if I could automatically create a transcript of the quote and then verify it's correctness? I know the technology exists, but what tools are available? For more about the technology, see Wikipedia's Natural-language processing:-

Natural-language processing (NLP) is an area of computer science and artificial intelligence concerned with the interactions between computers and human (natural) languages, in particular how to program computers to process and analyze large amounts of natural language data.

For more about available speech-to-text tools, I'll have to make a survey and actually try some of them.

11 June 2018

The Lineage of Leela

And AlphaGo Master begat AlphaGo Zero, and AlphaGo Zero begat AlphaZero, and AlphaZero begat Leela -- at least according to The Lineage of AlphaZero (January 2018) and Understanding Leela (June 2018). And behind all of this begetting was David Silver of Google's DeepMind.


Deepmind AlphaZero - Mastering Games Without Human Knowledge (42:29) • 'Published on Jan 29, 2018'

From Youtube's 'The Artificial Intelligence Channel'

2017 NIPS Keynote by DeepMind's David Silver. Dr. Silver leads the reinforcement learning research group at DeepMind and is lead researcher on AlphaGo. He graduated from Cambridge University in 1997 with the Addison-Wesley award. • Recorded: December 6th, 2017

If you're as out of the loop as I am, you might like to know that NIPS is shorthand for 'Conference and Workshop on Neural Information Processing Systems'.

10 June 2018

A Smashing Game

Have you ever felt like smashing your chess set with a hammer? How about smashing your board? These guys did and they weren't even angry -- maybe just a little mad.


What's inside a Magic Chess Board? (7:26) • 'Published on May 13, 2018'

The description started,

You've never seen a chess board like this - It plays against you! We have to figure out how this thing works!

And it ended,

WARNING: Cutting things open is dangerous. We do not recommend you try what we do! [...] Remember our motto "We cut things open so you don't have to", so we do not recommend you try anything we do.

The board is from Square Off ('Connecting the world on a telerobotic chessboard').

08 June 2018

An NN for Chess Images?

After getting my introduction to AI in It's Much More than Chess, can I make any practical use of the knowledge? I'm already following the chess playing aspect -- as in Understanding Leela -- but there must be other angles involving chess.

In yesterday's post Chess in The Graphic, I mentioned my archive of chess images. Maybe I could do something with that? For starters, how about building and training an NN to tell the difference between a photo and a drawing; and maybe throw in a scanned text as a third possibility. Take the following image as an example.

The accompanying text says,

This is a fantastic 20"x24" hand painted, unstretched oil on canvas. The painting depicts two dalmation dogs playing chess. The oil is unsigned.

Could a network recognize that this is not a photo? Or (without being told) that it shows dogs. My first step should probably be to make some sort of survey of what software and services are available.

07 June 2018

Chess in The Graphic

In a post from last week, An 1886 Photoshopped Illustration, I featured,

"The Sixteen Leading Chess Players of the World" from the July 17, 1886 issue of The Graphic. [...] I found three copies of the image in my archive. The best copy showed that the 1886 original had the names and nationalities of the players listed outside the border of the image. The best associated text copied that information.

While I was browsing the archive I found nearly a dozen other images from The Graphic. Here are four of them.

Legend (UL = Upper Left, etc.):-

  • UL: 1890-08-23 - 'The International Chess Congress - Some of the Chief Members of the Liverpool Chess Club'
  • UR: 1883-02-10 - 'Chess with Living Pieces - A Match at the Guildhall, Winchester'
  • LL: 1883-05-19 - 'A Military Chess Tournament with Living Pieces - Capture of the Bishop'
  • LR: 1883-10-27 - 'A Chess Tournament with Living Pieces at Brighton'

The fellow in the foreground of the first image is J.H.Blackburne, apparently conducting a blindfold simul. As for the three 'Living Pieces' images, I hope my legend is correct, because I discovered some contradictory text attached to the different images.

I could have resolved these discrepancies with digital copies of the original issues of The Graphic. Here, for example, is a link to the British Newspaper Archive ('in partnership with the British Library') showing the issue containing the image from '1886 Photoshopped Illustration': Results | From 17th Jul 1886 | The Graphic. It's a subscription site and I didn't want to pay just to resolve some minor issues, so I'll leave that for a time when I have more serious research to do.

05 June 2018

June 1968 'On the Cover'

Fifty years ago the covers of the two leading American chess magazines featured the nation's premier open tournament and the world's premier qualifying tournament. For the previous month's covers, see May 1968 'On the Cover'.


Left: 'Maroon Bells, Colorado; See back cover for U.S.Open details'
Right: 'Spassky: 1966 Challenger on His Way Back?'

Chess Life

[Back page ad] 69th Annual U.S. Open Chess Championship • Twelve Rounds -- In the West Village Conference Center • Snowmass-at-Aspen, Colo. -- August 11-23 • Tournament Director: International Master George Koltanowski • Assistant Tournament Director: Col. Paul Webb • $4500 Guaranteed prize fund • 1st Prize $1250 plus trophy [...] • Entry Fee: $25 plus USCF membership if not currently a member

The first time on this blog that we saw the venue of the U.S. Open was July 1964 'On the Cover' ('On to Boston!'). This was followed by 'On the Cover' for May 1965 ('A Caribbean Chess Holiday!'), and May 1967 (Atlanta). In 1966, the cover for the Open was usurped by other tournaments.

Chess Review

Boris Spassky of the Soviet Union is leading the prospective challengers in the elimination matches.

The magazine continued, 'The first round of Challengers Matches has probably ended.' Of the four quarterfinal matches for the 1967-69 Candidates Matches, Larsen - Portisch (+2-2=3 each) was still unfinished at the time the June issue was published.

04 June 2018

Understanding Leela

I ended my previous post Running Leela, with an action:-

Suddenly Black played 34...Qf4, placing the Queen en prise to two pieces. Give that move '??'. I don't understand what happened here and will look into it another time.

First stop: LCZero - Google Groups is an active discussion group where the participants are keen Leela fans with varying knowledge. It didn't take long to find a thread -- What is going on here? (May 29) -- which started with the same question I was asking:-

The above game contains a lot of seemingly absurd moves. Does anybody know why this is happening?

The answer had two parts:-

Training games have randomly selected moves to enable learning. If you want to analyse games played between networks please look for them here: lczero.org/matches.

So training games are the first step in the Leela heirarchy; matches are a second step. Got it. The same post pointed to another introductory resource: leela-chess/wiki/FAQ (github.com).

[Self-play training games] are how Leela Chess Zero trains herself. They are played with extra randomness turned on so it can discover new good (and bad) moves. This means the quality of these games is lower than the match games.

The FAQ should be a good place to start the next time I have a question. Back to the discussion group, how to discover the most relevant threads? I started by looking at recent threads with a high number of responses. Here a few entry level examples:-

  • Delayed gratification, 50-moves rule, discounted rewards, and an exercise (May 21; 73 posts by 14 authors) • 'There are earlier discussions on the LCZero group about Leela Zero aimlessly moving about in a won position, apparently constrained only by the 50-move rule to preserve its anticipation of a win. I don’t like this behaviour, but I have seen it defended by appeal to Zero Knowledge principles; a win is a win and we don’t care for elegance. I like to make a point here, backed by higher authority, that we should care for elegance and that it is appropriate to favour quicker gratification'

  • Super weird match from training games (May 26; 38 posts by 7 authors) • 'lczero.org/game/14801114; Super good start, but from move 29 and anything beyond, this looks like solid randomness.' • 'I wanted to reply that training games are irrelevant, but yeah this doesn't look right. It almost looks like both sides were trying not to win.'

  • Ideas to increase the interest in developing Leela Zero among the general public (May 31; 20 posts by 16 authors) • 'I have been following Leela Zero project for a couple on months now. I am a chess player with little knowledge in AI. I have noticed the number of training games per day is going down, which is going to make the development even slower.'

There are also many technical threads that discuss different configurations for the Leela client or that delve into the structure of the network(s). Let's look at those another time.

03 June 2018

Caveat eBay Digital Documents

Starting with the previous edition of Top eBay Chess Items by Price, I switched the long-running series (first post March 2010) from a fortnightly to a monthly schedule. Since that previous post -- Chess Art, Paris 1925 -- appeared at the beginning of May, I had eBay auctions from the entire month of May to choose from. Did that mean I had double the number of interesting items on my short list? Not exactly.

Of course, there are always auctions for collector chess sets or chess computers, but they have to be especially unusual to catch my attention. In a decidedly uninspiring month, the item pictured below grabbed the lion's share of that attention. Titled 'Chess Life and Review - 2 disk DVD collection Chess Software', subtitled 'Brand New', it sold for US $582.04 ('+$19.60 shipping'), Buy-It-Now.

Back in 2012/2013 I downloaded these documents from the web and they have since served as inspiration for many posts on this blog. See, for example, the most recent 'On the Cover', for May 1968; the 2017 series on Early U.S. Ratings; or an overview of the CL/CR digital copies, Shaping Chess History (September 2016). The description of the item pictured above started,

A must have item for any serious chess historian or chess collector! This 2 DVD collection contains digital reproductions of every issue of...

In fact, there is no need for me to copy the description, because you can find it yourself on Amazon.com (where the DVDs currently sell for $43.94) and USCFsales.com ($39.95). Just use the key phrases from the quote I've given. All three descriptions also use the sentence...

Here is your chance to get 42 years worth of Chess Magazines at a cost of less than $1.00 per year!

...which is obviously not true for the eBay item. The seller of that item has more than 100 'Neutral' or 'Negative' feedback comments, some of which mention that an item sold on eBay was shipped using Amazon Prime. Like many once-interesting web ideas, eBay eventually evolved into a giant swamp. The chess DVDs just add to the widespread muck.

By coincidence, while I was preparing the recent post, An 1886 Photoshopped Illustration, where I mentioned that 'I've been downloading old copies of The Chess Journalist (TCJ)', I noticed that the December 2006 issue of the TCJ credited the existence of the scanned CL/CRs to Tim Tobiason. He seems to have been a colorful character in several ways, but this isn't the time or place to repeat stories that can be found elsewhere on the web. It is his misfortune that while the original magazines are protected by copyright, his scans aren't protected by a second copyright because they don't represent creative work.

***

Later: Re that last paragraph...

He seems to have been a colorful character in several ways, but this isn't the time or place to repeat stories that can be found elsewhere on the web.

...the Armchair Warrior found both the time and the place: Tim Tobiason - A Colorful Chess Character (xpertchesslessons.wordpress.com).

01 June 2018

It's Much More than Chess

Thanks to a post I wrote at the end of last year, The Constellation of AlphaZero (December 2017), I became curious to know more about the technology used in AlphaZero:-

The excerpt I copied from the paper also talks about parameters. It ends, 'The updated parameters are used in subsequent games of self-play.' I wonder if I can find out more about those parameters. I'd also like to know how all of this was used in the match that crushed Stockfish.

This led me to a series of courses on Coursera called Deep Learning (coursera.org):-

In five courses, you will learn the foundations of Deep Learning, understand how to build neural networks, and learn how to lead successful machine learning projects. You will learn about Convolutional networks, RNNs, LSTM, Adam, Dropout, BatchNorm, Xavier/He initialization, and more. You will work on case studies from health care, autonomous driving, sign language reading, music generation, and natural language processing. You will master not only the theory, but also see how it is applied in industry. You will practice all these ideas in Python and in TensorFlow, which we will teach.

I had already taken a bitcoin/blockchain course on the same site and knew that their material was reliable. It took me about five months to work through the five courses, which were titled:-

Neural Networks and Deep Learning
Improving Deep Neural Networks: Hyperparameter tuning...
Structuring Machine Learning Projects
Convolutional Neural Networks
Sequence Models

One of the big advantages of Coursera is that the courses can be audited for free. Although all material for the courses -- videos, quizzes, programming assignments, and forums -- is available at no cost, the quizzes aren't graded without payment and there is no certificate issued for completing a course. Not having done any programming in more than 20 years, I was concerned that my skills were out of date, but there was plenty of support offered for both Python and TensorFlow (Keras is also used) and I had no trouble completing the assignments. This was partly due to the cookie-cutter structure of the exercises and to forum discussions by previous students that shed sufficient light on the knottiest problems. It also helped to know some math, particularly linear algebra and calculus.

The courses are the work of Andrew Ng (wikipedia.org) and his company, Deeplearning.ai. After noting a few weeks ago that AI and all it entails is A Transformational Technology (May 2018), I'm glad I understand the subject much better than I did six months ago.

A key question I had going into this endeavor was 'Why is AI taking off now?' People have been talking about it for decades, but it was always just around the corner. What changed to make it reality?

An important driver has been 'big data'. We now have so much digital data with huge portions of it correctly cataloged and labeled that we have the means to explore digital relationships that were inaccessible in the past.

We also have algorithms to process big data that, not so long ago, were unknown or undeveloped. Most modern AI is based on the same set of related algorithms. Computing hardware has progressed in parallel with the development of those algorithms allowing for their practical implementation. The necessary support for vectorization in both processors and software development tools was previously missing.

Finally, the impetus for capturing big data and developing big tools is big money. AI first revolutionized the commercial endeavor of selling ads and is now gradually creeping into every commercial activity on the planet.