Friday, June 26, 2015

Interfaces

Many have tried but only a few succeed. It isn't easy to transform your hobby into a profession especially if this hobby is chess. Besides your chances drastically diminish when you aren't a worldclass-player. Chesscafe is the most recent example of failures. Recently they switched from free to paying but not even a year later they had to announce a hiatus. It is highly confrontational to discover that almost 20 years of free publications don't guarantee anything commercially.

On this blog I have been regularly criticizing Chessbase (e.g in the article misinformation by chessbase) , but it is beyond any doubts that the company is the most successful one in our chess-domain. A very clever marketing-strategy assured the company more or less a monopoly. They were one of the very first ones to see the connection between frequently reporting about chess-news and selling their own products.

Almost a decade ago Chessvibes (today Chess.com) started to compete reporting chess-news. Initially Chessbase didn't care much. However this changed when they started to feel the consequences of not being anymore market-leader in chess-journalism. To protect the sales of their products they had to improve their model. They introduced finally the option to give direct feedback below the articles instead of the old fashioned, very limited mechanism of emails. However another important change was the involvement of (cheap) young (grand-)masters to comment games like Alejandro Ramirez (Costa Rica) and Shah Sagar (India).

Chessbase clearly has more financial reserves than their competitors which is reflected by their investments. One investment they made is the Let's check development. Let's check is nothing else than a gigantic database of all the collected positions by Chessbase with an evaluation made by an engine. The brilliance of the concept is that all the analysis are made by the users of Let's check while simultaneously the users will have to pay to consult their own analysis in the future. Chessbase only had to develop the interface and make sure enough storage is foreseen for the database (which can become a headache in the future with the large speed the system is growing).

I am astonished how fast this database is growing. The mechanism of awarding points seems to be a very good carrot for quite some users. When Chessbase launched the system I was chuckling but now only 4 years later I have to admit that I am blown away by the amount of data available many times bigger than whatever opening-book. I was pretty shocked recently to find out that a completely insignificant position from my clubchampionship-game against Marcel (see catenaccio) was already analyzed at a depth of 38 plies.
Let's check
Today consulting Let's check has become a fixed part of my analysis-routines. A negative remark which I heard is that the old analysis automatically slowly disappear out of the database. Personally I think this is something positive as it is essential that the analysis are made by the most recent and strongest engines otherwise the value of the database quickly decreases. Besides somebody will more quickly be tempted to connect an engine to the database when few or no analysis are mentioned. Finally it is also a mechanism to keep the size of the database under control. I recommend people interested to know more details about the tool to read the online review from HK5000 (Dutch language).

Although today I am fully convinced about the added value of the Let's check feature, it took me quite a while before I wanted to give it a try. It was only when I bought a couple of months ago Komodo via Chessbase instead of directly from the developers that I finally got access to all the features. Maybe 1 of the better reviews about Komodo has been written on chessbookreviews. I want to add that today I see very little difference between evaluations of Komodo and Stockfish 6 which is pretty much aligned with the result of the final of the 7th TCEC season. On top I also noticed that using very low timecontrols (10 seconds or less per move) that Komodo more often misses something tactically which again was partly confirmed by the CCRL 40/4 rating-list (40 moves in 4 minutes) which Stockfish leads in contrast to CCRL 40/40 rating-list (40 moves in 40 minutes). So if you are not interested in making very deep and extensive analysis then the free engine Stockfish is more than sufficient.

This is something about which Chessbase is also aware of course which explains their strategy to add some nice extras to anyway convince you buying their product. Except Let's check you also get 6 months premium subscription to Playchess and you get the new Fritz 14 interface. The added value of the subscription is evident but it is not clear what to expect from the interface.

Last couple of years I worked with the Fritz 11 interface which was ok for me. Besides a new interface always demands adjustments as some shortcuts change which I like to use (see e.g. using databases). In the beginning this causes a lot of annoyance when I once again use the old shortcut wrongly shutting down the program without saving the analysis. It neither is helpful that Fritz 14 has no manual. Reference is made to the manual of the Fritz 13 interface but I already detected several differences or I still miss certain details. The buyer is a bit left alone.

However there is one thing which let you forget all the previous obstacles. The interface clearly increases the strength of the engines. I compare the same analysis for both interfaces. First a screenshot with the Fritz 11 interface.
Fritz 11 interface
Hereafter a screenshot of the Fritz 14 interface.
Frit 14 interface
Same computer, same engine, same position and same period and still we see a serious difference between both analysis. The version 14 is almost 3 times faster as the version 11 which even gives us a gain of 3 plies. Well don't pin on the 3 plies as surely Stockfish prunes enormously in the tree of variations but undeniable we see a considerable rating-gain. 70 elo seems to be the difference between 1 core and 4 cores for Stockfish looking to CCRL 40/40  in which 1 core corresponds to the Fritz 11 interface while the 4 cores is the Fritz 14 interface.

We are very eager to acquire a new version of an engine but the interface is sometimes forgotten. Komodo 9 wins only 24 ratingpoints compared with version 8 conform the most recent tests. I will surely better follow up new developments on the interfaces in the future.

Brabo

No comments:

Post a Comment