Questions by Roger Brown, Christopher Conkie, Michael Diosi, Thomas Hall & Alex Schmidt

Hello Mark. Thank you for agreeing to give us this interview.


Question 1.

If you ask anyone who uses Hiarcs why it is one of their favourite engines the answer is nearly always the same, it’s the human like playing style. Which human chess player would you compare most with Hiarcs or is there even a determined player you try to emulate? And besides that, with which player are you personally most impressed?

I am not sure I ever had any one player in mind. Certainly I think Hiarcs plays in its own style. I admire many great chess players and the ones who have influenced Hiarcs' development most would be Tal, Kortchnoi, Kasparov and Anand.

Question 2.

We notice that you do not release a new version of your engine on an annual cycle as do other authors. Why is this?

I believe a new version should only be released when it is clearly better and probably at least 40 Elo stronger than its predecessor. I have been following this standard since HIARCS first become a commercial chess program nearly 15 years ago. So it is ready when it is ready.

Sadly, we can see from the rating lists many other programs are released even when it is debateable whether they are any stronger. Part of the problem is that the market for strong chess programs is tiny and the remuneration likewise. Therefore it becomes very difficult to make money unless the program strength can be continuously improved. So I understand in a way why there can be the need to release in time for Christmas etc irrespective of other factors.

Question 3.

What is the thinking, the motivation behind Hiarcs 10? What inspires you to write the code that you do?

I am motivated by creating the strongest chess playing entity with a very entertaining style of play. To me this also means encapsulating as much chess knowledge I can without adversely affecting the chess strength.

Question 4.

What are the new/updated and improved features in the newly released Hiarcs 10 engine?

I'll cheat here and grab the advertising blurb about Hiarcs 10: "Hiarcs has long been renowned for its human-like playing style, now Hiarcs 10 goes much further with improved chess knowledge, a more aggressive style and an ability to search very deeply for the truth of a position. New chess knowledge enables Hiarcs 10 to identify deep attacking motifs long before they become apparent especially concerning king attacks and unbalanced positions and material. Hiarcs 10 often prefers initiative and attacking options over more materialistic concerns associated with traditional chess programs. Matched to this new chess knowledge Hiarcs 10 searches much more deeply than its predecessor often enabling it to find new unexpected moves and ideas much faster."

Some of these attributes can be seen in some selected SSDF games which can be played through online here: http://www.hiarcs.com/Games/H10games.htm

Question 5.

Recently Hiarcs played Polish Grandmaster Piotr Bobras in a rapid match and won 3-1. Are there any more GM matches planned for Hiarcs in the near future?

That was Palm Hiarcs which played that match courtesy of Michael Enderle who has played some wonderful matches with Palm Hiarcs online. Actually a lot of these games and some of the SSDF games of Palm Hiarcs can be played through online here: http://www.hiarcs.com/phresults.htm including the great match victory against GM Jan Gustafson at G/30+10.

Currently we do not have any other matches planned. However we have a number of GMs very interested in the new attacking capabilities of Hiarcs and we hope to work with these GMs to improve Hiarcs.

Question 6.

Will there be a multiprocessor version of your engine? (Maybe there is already)

Yes a multiprocessor version of Hiarcs is planned.

Question 7.

Do you personally play chess regularly and if so where and at what level?

Yes I play for a new local club which was only formed a few years ago. We started in the Hertfordshire chess league division 6 and have won our division in each of the last four years getting promoted each time. Now in division 2 we are currently leading the division but I do not know if we are strong enough to win the division and get promoted to the first division, we will see.

Question 8.

How do you feel about the use of opening books and endgame databases in computer chess?

I think they are very useful! It is interesting to work on the opening book and I enjoy that work. I work with Eric Hallsworth on the opening book.

Endgame databases have improved the analytical ability of chess programs in the endgame, but I am not convinced they improve the chess strength very much. In some ways it is a pity as many authors now do not work on late chess endings and some programs cannot even mate with B+N without tablebases!

Question 9.

You always stated your primary target is to improve Hiarcs’ chess knowledge and we all know the abilities of Hiarcs regarding positional play. Is it actually possible to improve Hiarcs by adding even more chess knowledge or wouldn’t it be more promising now to improve the search and the tactics?

It is very possible to improve the chess knowledge of any chess program including Hiarcs. I worked on the search in Hiarcs 9 and Hiarcs 10 and made good improvements there. I believe a chess program should improve in both chess knowledge and its application in the search between each release and I continue to try to achieve that.

Question 10.

The computer chess scene has been revitalized by many new strong engines recently. Some are heavily knowledge based like Hiarcs. When programming knowledge in an engine what would you say are the major concepts that a programmer should look at?

I don't know of any others which are so heavily knowledge based but there are certainly strong engines around. There have been a number of new strong engines and we have Fabien and Tord to thank for their significant contribution to the chess computer community.

I think it is important to understand chess, read good chess books on positional play and play chess yourself. Often it is difficult to identify, isolate and extract the essence of the chess knowledge all GMs know but often cannot adequately explain in way we as programmers can codify as chess knowledge efficiently or effectively. Its also dangerous because bad knowledge is much worse than no knowledge! Perhaps this is why many programmers prefer to concentrate on search and it is surprising how search is able to substitute for knowledge and give the impression of improved chess knowledge.

Does chess knowledge simulate search or does search simulate chess knowledge?

Often overlooked here is the fact that just as important as the chess knowledge is the balance and this is often difficult to achieve. Lastly, test test and test.

Question 11.

Do you think that your approach is somehow indicative of a search for truth in chess if such a thing exists? By this I mean do you think that Hiarcs will discover a way to approach the game that will have persons sighing in amazement?

I think the approach is to try and play interesting, entertaining chess and hopefully win as many games as possible. I can also say this is not easy to do.

Question 12.

Hiarcs is clearly more than just about ruthless speed or nodes per second. Why is knowledge so important to you? Why not go after those other things that can be measured by the average user?

I would be bored stiff if I had to just write a fast searcher. I need to be interested in the development. Is chess knowledge not important? is it important to others? I think it must be. What is chess knowledge anyway? What things are you referring to which can be measured by average users? Nodes per second or something else? Bean counting hardly seems important or interesting to a chess player. Its a bit like just looking at numerical match results and not looking at the chess played. Its easy to fall into this and miss the point!

Question 13.

Which is the primary program language that you use to program Hiarcs and what are the reasons why?

I use the C language because it's efficient and comes naturally to me.

Question 14.

Standard alpha-beta algorithms concentrate chess knowledge in the evaluation routines, are you doing this also or is it different in your chess engines?

I use alpha-beta yes. I do believe the chess knowledge needs to be spread throughout the program and not just in the evaluation function. In a way the underlying ethos in Hiarcs is different from a traditional chess engine design but it still has a lot in common too.

Question 15.

At what point are you satisfied with Hiarcs that you decide to release it?

As I said before, it needs to be much stronger and it needs to have the right balance so it plays interesting entertaining chess. I also set a 40 Elo hurdle to jump!

Question 16.

Do you think that one day you will consider writing a UCI version of your engine so the users out there will be able to enjoy your engine in several GUI’s?

I have a UCI HIARCS chess engine product for Macintosh which works well with Sigma Chess and that has been available since January 2005. See http://www.hiarcs.com for details. I also have a PC UCI HIARCS chess engine running supporting its own opening book. I hope to release this as a product in the near future.

Question 17.

Do you think that computing power (hardware strength) will play the decisive role in the future of chess engines?

Hardware is important but so is software and both play a critical role of course. At present the software is advancing as fast or faster than the hardware.

Question 18.

What is the long term future for your engine? Do you wish it to dominate the engine-engine scene (SSDF lists etc) or do you want it known as GM killer? Or are perhaps these goals are of no interest to you?

My goal is to produce the strongest chess program which plays interesting, entertaining and challenging chess. It is of course very nice to lead important rating lists such as the SSDF list. Hiarcs also has a good reputation with some of the top GMs and I would like to build on that.

Question 19.

Unfortunately Hiarcs did not appear at the IPCCC Paderborn tournament in December 2005 which is just past. Which is the next computer chess tournament we will see Hiarcs appear in?

At present I am not sure, we will see! :-)

Question 20.

And finally the most important question of all (maybe). Who do you think will win the Football World Cup held in Germany later this year?

Compared to chess such things are of little importance, but of course it will be interesting to see if England can win the World Cup again. :-)

We would like to take this opportunity to wish you continued success with Hiarcs.

Thank you!