Keeping your scoresheets

[Andrew Craig]

I've kept all mine since returning to playing chess about 10 years ago.

I actually digitised all my games in the past couple of months whilst between jobs - 'only' 200 in total - I don't want/won't have time to do that again! Have backed up and also saved my database in a large pgn file (which is sync'd in the cloud), which was a tip I took from one of our club members.

I've been long wondering if someone would develop a program to digitise via a scoresheet - thanks for the link Ian Thompson - I tend to make notation errors but will have to give it a test and see how successful it is at understanding my scrawlings.

Sadly my junior games are all gone as a result of parents moving house and my having stopped playing. Not that I expect the quality was anything to write home about, but lost to history is my 4-5 move combination mating net against Jonathan Swindells to clinch the U13 Northumberland County Championships

[John Saunders]

The problem with any app of this type is that you need it to be 100% accurate; 99% accurate isn't good enough unless you're prepared to manually check everything it did which rather defeats the purpose of the app.

Yes, that's right. OCR can never be 100%, not even if a score is clearly and accurately printed. I remember trying to scan and copy games from a bulletin which used KQRBN (i.e. non-figurine) algebraic notation and it seemed to work reasonably well, with only a few manual amendments being needed. I might have continued to use it had that been the limit of the problems encountered but it later transpired that quite a number of the scores were wrong because it had confused (for example) Rc2 with Re2. (As regards figurine notation it was a non-starter as the OCR software I use, ABBYY Finereader 12, cannot be 'taught' to recognise them accurately, although in theory it is supposed to be possible to 'teach' it to do so.) In the end I concluded that manual input was probably quicker anyway as it allows you to check for errors as you go along, including errors made by the player filling in the score, of which there are many, particularly in the run-up to a time control. A skilled game inputter can often figure these out, and also guess moves where the player has left gaps, taking into account the standard of the players involved. Could a clever app of the future do as well? I have my doubts.

[Jonathan Rogers]

odd as it may seem, this reminds me of my game against Paul Byway at Eastbourne 1991(1) which I am delighted to find on John's (wonderful) site https://www.saund.co.uk/britbase/pgn/19 ... iewer.html

Here I played 22 Rb1, and not 22 Rf1, as given in the bulletin, and therefore also in the game on Britbase. I daresay that my scribbled Rb1 could have looked like Rf1, and I can see that the game inputter at the time might have thought it plausible enough so as not to question it.

I remember this because I noticed this at the time (there were daily bulletins after each round) and so the next day I asked Byron Jacobs in the bulletin office whether it could be corrected, at least in the final version at the end of the event. My impression that he was quite uninterested in doing so proved to be well founded, and so the error lives on in the official records... not, I hasten to admit, that anyone has, as far as I know, found the game online and criticised me for playing it!

[John Saunders]

Thanks, Jonathan. I've amended the game on BritBase and also alerted chessgames dot com - they usually amend promptly.

[Paul Habershon]

Thoughts and facts about my score sheets.
1) I don't keep any unless I have played someone famous (e.g. one signed by Tigran Petrosian after a simul).
2) Echoing Graham Borrowdale's 'sad' comment, I do copy all into scorebooks. I was glad to discover recently that an English GM still does the same.
3) I started the practice in the year after John Townsend (above) on 1st January 1962 at my first congress, Hastings, in a New Year Special.
Although I must have played a few school matches previously I think I wasn't even aware that score books existed. There were BH Wood type 'Chess' score books on the bookstall.
4) My first few books were similar, but from 1978 I started using up a bulk buy I got at a BCF Congress. The intention was to sell them on to pupils in my school chess club, but I somewhat overestimated demand. The seller, with his name and business address on the scorebook cover, was a notorious British Champion from Bolton-on Dearne, Rotherham. I am now on my 58th book of 50 games and I have enough of those books to see me out.
5) Unlike Roger I am far too lazy to use my games for self-improvement. However, I have sometimes found them useful for historical contributions, particularly to this forum.
6) I regret not being as diligent with my correspondence games. I am currently trying to clear my loft and have already found several from a period in the British Correspondence Chess Championship cycle (Reserves, Candidates and Championship sections). I am hoping to find a note from Jim Slater (financial saviour of the Fischer/Spassky match). He withdrew after a few moves, presumably too busy with his burgeoning career.
7) Should I instruct my family members who will eventually have to sort out my personal effects? I have no idea. I will probably say nothing and let them consign the score books to the skip.

[Jonathan Rogers]

Thanks, Jonathan. I've amended the game on BritBase and also alerted chessgames dot com - they usually amend promptly.

Thank you too! (That includes your maintenance of the whole site).

[John Townsend]

Paul, in one of my books I find our game of 18 March 1972, Beds. v. Berks.: an attacking game by both sides, but also with at least one very bad move on either side!

[Kevin Thurlow]

Initially, about 60 years ago, I recorded games straight into scorebooks (when you could do that), but errors in notation were a problem, so I used a scoresheet and copied it into a book. I got Chessbase 3 (I think) in the mid 80s and put the game on there first, then copied it into the scorebook, correcting the mistakes. Gradually, I updated the database with previous games. At some stage I started recording the time taken for each move, but put that in the scorebook and didn't bother in the database. I did find a box of scoresheets from RP events and have put those on Chessbase. I have a different set of scorepads for Correspondence chess, but the games go on Chessbase. Online games I keep in a separate database, mainly as they are not real chess.

One problem is that old Chessbase only had they year for the game, so I need to update the exact dates of games. Also the event was frequently given as something like "R1" or "SyIII2", rather than "Redhill vs X board 1", or "Surrey III vs Y board 2", so I hope to work out who the opposition were playing for. Also, I tended only to use initials in the old days, whereas now the databases and I tend to use more detailed names. The well-known chant of "There's only one Tigran Petrosian" no longer applies.

And I must add my admiration of John Saunders and his fantastic work. I sympathise with Jonathan's attempt to correct a typo. One of my games at Guernsey appeared on a site which purported to give a world ranking based on the small number of games they had. I beat a 1900 in the last round many years ago, but the bored junior that was press-ganged into inputting games there managed to get the result the wrong way round, and the opponent's name wrong, so it appeared I had lost to a 1300. (I wasn't taking the world ranking seriously of course...)

Edit - I meant to add that whenever I look at my database, I tend to get distracted by old games and randomly look at them, especially when I had forgotten I'd played someone!

[Matt Mackenzie]

Still copy the past season's efforts into a scorebook every summer - have a complete record going back to September 1981 as a result.

Originally kept my junior and quickplay scores, but stopped that in the mid-80s and then "purged" them from the scorebooks which I started doing around then - something I now regret as they contained some interesting encounters. Still, getting on for 1200 "classical" games even so.

[Paul Cooksey]

I have a chessbase file. Am I a heretic?

[Nick Ivell]

I have about 1000 sheets from 1973 to 2020. The pandemic saw me off (chess-wise I hasten to add).

They started off nice and neat but descended into Fischer-like illegibility towards the end.

I hardly ever use them but I regard them as valuable souvenirs of some of the best moments of my life. I could never be bothered to transcribe them into a scorebook.

[Roger de Coverly]

One problem is that old Chessbase only had they year for the game, so I need to update the exact dates of games. Also the event was frequently given as something like "R1" or "SyIII2", rather than "Redhill vs X board 1", or "Surrey III vs Y board 2", so I hope to work out who the opposition were playing for.

I have a similar problem. It's fine for the years where there's a written scorebook, but once I abandoned those, the detail is missing. I did however figure out that if I had an archive of diaries for the period, I would be able to reverse engineer the dates. For SCCU county matches, the SCCU Bulletin archive would supply dates.

[Paul Habershon]

Paul, in one of my books I find our game of 18 March 1972, Beds. v. Berks.: an attacking game by both sides, but also with at least one very bad move on either side!

John, you modestly refrain from mentioning that you won that game. Having just put it through an engine for the first time, I have to admit that my play was a disaster from which I have learned mainly that I have much to be modest about!
Anyway, for me this kind of exchange is what keeping score books is all about.

[Roger de Coverly]

Yes, that's right. OCR can never be 100%, not even if a score is clearly and accurately printed. I remember trying to scan and copy games from a bulletin which used KQRBN (i.e. non-figurine) algebraic notation and it seemed to work reasonably well, with only a few manual amendments being needed. I might have continued to use it had that been the limit of the problems encountered but it later transpired that quite a number of the scores were wrong because it had confused (for example) Rc2 with Re2.

I wonder if anyone has managed to cross breed an OCR reader with a chess engine. Even a weak one would filter illegal moves and flag improbable ones. It's likely to work better the higher the standard of players whose games were being scanned.

[John Saunders]

I wonder if anyone has managed to cross breed an OCR reader with a chess engine. Even a weak one would filter illegal moves and flag improbable ones. It's likely to work better the higher the standard of players whose games were being scanned.

I've wondered that too, but I suspect it would still fall short of the required 100% accuracy, particularly with regard to the Re2/Rc2 syndrome, plus frequent disambiguation failures (often to be found in even the best produced bulletins in the pre-ChessBase era, e.g. Rc8, where both Rac8 and Rfc8 are plausible and subsequent play doesn't provide a solution). If we are talking about the input data being the original scoresheets, then dashes for moves made in time trouble or inaccurate post-game fill-ins of previously unrecorded moves are all too familiar. In extreme time trouble even GMs can play nonsense moves and this carries over into a post-game brain fade when the players may be hassled by arbiters to complete their scores and hand them in. One of the skills required by the inputter is to know when they are beaten and write "indecipherable score" at the appropriate juncture, but this has to be a last resort. The less skilled ones tend to resort to this rather too soon. Often the mistake is their own. Having input a duff move earlier in the game, they run into an apparent illegality, don't have the savvy to check through the whole score and give up. They are too quick to assume the mistake is the player's.

Game inputting has improved markedly since the 1990s, when the inputter had a heavier workload (pre-electronic boards) and the technology was generally less helpful, but the job still gets botched occasionally. I suspect game databases have a larger percentage of score errors than people realise. Many of the most important historical tournament games were input in the 1990s when CB software was in its infancy, and game inputters were inevitably less skilled than they now are. Rather than bringing out gimmicky new versions of their software, I think ChessBase should put some resource into a complete review of their databases for errors, adding round numbers and dates where they are missing, etc. If they are bent on updating the software, then they might want to think about providing some additional/enhanced features for comparing versions of scores, removing duplicates and sorting games, none of which their software does satisfactorily at the moment. (Or does it? I don't have the latest version but I've not seen enhancements to any of the above features for many a long year so suspect they have not been improved recently.)