I have now added a new page to the statistics section on Meson. It provides totals and percentages of unsound and anticipated directmate problems. The unsoundness indicator is just an entry in the classification XML, and counting up the unsound problems takes longer than most queries on Meson do. So, when you click on the 'Numbers concerning unsoundness and anticipation' link on the statistics page, don't be too surprised about the delay of a few seconds before the server sends the next page.
'AntCheck' is running again, looking for #2s which are 80% or more the same as each other. As I write, it has checked 24,500 of the problems and recorded 520 possible hits for me to investigate. As both sides of a 'match' normally show up during this process, that figure of 520 is probably not as bad as it sounds. To try and eliminate 'false positives' I have tightened up the comparison routines a bit, but only time will tell what difference, if any, that will make.
The number of unsound problems is worrying and I have started to investigate them, starting with the #2s in the FIDE Albums. So far I have been through the unsound #2s in the first two albums and found 3 where the error was mine, getting the diagrams wrong on entry to Meson. These have been corrected and I shall continue my enquiries.
Having mentioned the FIDE Albums, it is probably worth stating that all #2s in all the Albums up to and including 1995-1997 (and annexe) are in Meson. Starting from 1914-1944 all other types of interest to Meson are included up to 1986-1988 and annexe. I am currently working my way through the moremovers in the 1989-1991 album.
Outlook for the next period: expect anticipation numbers to rise and unsoundness numbers to drop!
Subscribe to:
Post Comments (Atom)
2 comments:
The description of 'complete anticipation' on the new page mentions only reflections and rotations. I assume there is an unstated 'or combinations of the two' intended.
But what about translations? The well-known Jokish / Laws 'twin' involves translation only.
I get the impression that anticipation is checked only at 'root' level -- but there are some cases of deliberate anticipation that are done on different levels, such as extracting a sound two-er out of an unsound five-er. I suppose the reverse operation is theoretically possible, but it seems to involve more work than plagiarists in general would be prepared to spend, and so would probably be more likely to indicate coincidences and subconcious influence.
That kind of anticipation checking, however, seems to require keeping track of every position in all solutions (including intended solutions ...). The question is, I suppose, if it would be worth it.
Anders
Your assumption is, of course, correct. If I remember, I will make an amendment to that page.
I, at least, will not be doing any checking of the 'non-root' level anticpations that you mention. Life is too short. However, for a WCSC a year or so ago, Peter Bakker and I composed a #5 'after' a #2. The earlier problem was acknowledged, but we didn't mention it was a #2 and at least one commentator can be forgiven for not checking and assuming that it was a #5. That problem may well be worthy of a posting some day, but not because it was a particularly good problem!
Post a Comment