Vonset???

[samson]

Dear friend,

Thanks for your attention. We word hard and make the new product lunch as soon as possible. Thanks again.

Best regards.
Samson

We all wait for a great new product

[samson]

Dear Greg,

Thanks for your question. Based on the excellent performance of Stockfish, we have basically confirmed that we will use the Stockfish engine in L6 Pro.
At the same time, our goal is to make L6 Pro connectable to the app to meet the teaching needs of teachers and allow more young chess players to share their playing process with others, especially teachers, at any time during the game. Thanks again.

Best regards.
Samson

Dear friend,

Our own company's technical team developed the engine by referring to existing algorithms.

Thank you for your attention.

Hi Samson,

Always interesting to have a new dedicated chess computer on the market. However, could you kindly elaborate on "referring to existing algorithms"?

Is this simply referencing Stockfish evaluations similar to the DGT Centaur?

Could you share the background/qualifications of your software team? Have they developed other chess engines prior?

Kind regards,
Greg

[Fernando]

So, we will have yet another chess computer executing Stockfish code? This is becoming somewhat boring. The only way to compensate for that is the following:
a) To offer a really good companion more or less in the fashion of those that adjust its level to the user by itself, gives advice, examines mistakes and in every sense are like e GM hired by you to assist you all the time.
b) To offer not only the main engine, Stockfish, but also the capability to connect with others engines, UCI way.
c) To put in the list of available engines of the computer old but interesting programs that were historically interesting and important, but now forgotten or not easily accessible, like Spracklen, Don Daily, Kittinger and Doenninger programs.

Even just ONE of those options would be welcome. I am not really sure anymore, now, that I will purchase a computer made by you just for running Stockfish.

Second thoughts regards
Fernando

[mclane]

If all products use Stockfish because it is for free, it is indeed boring. There are thousands of engines.

[samson]

Hi Fernando,

Currently, our team focuses more on user practical scenarios. Our main concern is how to apply these technologies to solve more meaningful practical problems in life with lower prices and better experiences. For example, the design of our three modes, the first mode is actually designed to encourage chess players to challenge opponents with higher levels to enhance their confidence in chess and achieve higher levels. Technically, such as engines, we currently only treat them as tools that can meet the needs of users, and we have not invested much in engine technology research and development. We are more inclined to be a product team dedicated to creating more and better cost-effective products to serve our users. Thank you for your professional advice, and we will take it to heart.

Best regards
Samson

So, we will have yet another chess computer executing Stockfish code? This is becoming somewhat boring. The only way to compensate for that is the following:
a) To offer a really good companion more or less in the fashion of those that adjust its level to the user by itself, gives advice, examines mistakes and in every sense are like e GM hired by you to assist you all the time.
b) To offer not only the main engine, Stockfish, but also the capability to connect with others engines, UCI way.
c) To put in the list of available engines of the computer old but interesting programs that were historically interesting and important, but now forgotten or not easily accessible, like Spracklen, Don Daily, Kittinger and Doenninger programs.

Even just ONE of those options would be welcome. I am not really sure anymore, now, that I will purchase a computer made by you just for running Stockfish.

Second thoughts regards
Fernando

[Fernando]

I understand that. You are not thinking of us, chess fans and collectors, but a lot more in the mass market. That's OK, but maybe some effort could be directed to satisfy us because, after all, we are a sizable chunk of the market and if you can please both average people and us, chess fans, that would be a successful strategy.

I do have not the numbers, but chess collectors and fans are a lot more than you think. This place is just one of many around the world. It is, besides, people that tend to be over 40 or 50, so people with some capacity to purchase. On the other hand, a mass-market typical customer probably does not look for L6 kind of products, but a lot cheaper ones. So, Samson, WE ARE THE GUYS you must please

My best
Fernando

[samson]

Dear Fernando,

Can you help list out the specific needs of a group of people like you for Smart Electronic International Chess? Please provide a detailed list for us. We will definitely consider your thoughts and opinions and are very willing to hear your feedback.

Best regards
Samson

I understand that. You are not thinking of us, chess fans and collectors, but a lot more in the mass market. That's OK, but maybe some effort could be directed to satisfy us because, after all, we are a sizable chunk of the market and if you can please both average people and us, chess fans, that would be a successful strategy.

I do have not the numbers, but chess collectors and fans are a lot more than you think. This place is just one of many around the world. It is, besides, people that tend to be over 40 or 50, so people with some capacity to purchase. On the other hand, a mass-market typical customer probably does not look for L6 kind of products, but a lot cheaper ones. So, Samson, WE ARE THE GUYS you must please

My best
Fernando

[Fernando]

Ok.
Here is the list. Surely coincides with what my pals here and elsewhere want, but probably there are other features I miss.
So, let's go to it:

1.- Not one but several engines embedded in the machine as options OR an easy way through the Internet to get engines working with the UCi protocol to load and play.

2.- A window capable of giving info about the line the engine is thinking, the eval, the ply or the depth of the search.

3.- A way to upgrade easily the main engine via the Internet.

4.- A good opening book with at least 10 thousand lines

5.- Option to put off or On the lights in the squares

6.- For fun, written humorous comments coming in the windows respect the game that is going on, Fritz style. It is not that difficult; you just associate a number of situations - blunders committed by the human side, for example- - with several lines of commentary, kind of "man, have you lost your mind?" or anything you consider funny or even mordant.

7.- The most possible amount of levels for all tastes. Example: 40 moves in 15 minutes, 40 in 30, in 60, in 90, and so on.

8.- A function to set a degree of variety to the moves selected by the machine. Or always the best move in the search or at random one of the good ones if they exist.

I hope this is complete, but not sure. Even if only one or two of these suggestions could be implemented, it would be great.

Fernando

[samson]

Dear Fernando,

Thank you very much. I have taken very seriously any points you have made about any products and technologies, and while I cannot guarantee immediate application to the next generation of products, I can express that our team is very open to any feedback. Your opinions are very valuable and insightful, and I am grateful for our encounter.

Best Regards
Samson


Copy to Clipboard

Ok.
Here is the list. Surely coincides with what my pals here and elsewhere want, but probably there are other features I miss.
So, let's go to it:

1.- Not one but several engines embedded in the machine as options OR an easy way through the Internet to get engines working with the UCi protocol to load and play.

2.- A window capable of giving info about the line the engine is thinking, the eval, the ply or the depth of the search.

3.- A way to upgrade easily the main engine via the Internet.

4.- A good opening book with at least 10 thousand lines

5.- Option to put off or On the lights in the squares

6.- For fun, written humorous comments coming in the windows respect the game that is going on, Fritz style. It is not that difficult; you just associate a number of situations - blunders committed by the human side, for example- - with several lines of commentary, kind of "man, have you lost your mind?" or anything you consider funny or even mordant.

7.- The most possible amount of levels for all tastes. Example: 40 moves in 15 minutes, 40 in 30, in 60, in 90, and so on.

8.- A function to set a degree of variety to the moves selected by the machine. Or always the best move in the search or at random one of the good ones if they exist.

I hope this is complete, but not sure. Even if only one or two of these suggestions could be implemented, it would be great.

Fernando

[Tibono2]

Hello dear computer chess enthusiasts,
Getting back with first insights after my Vonset L6 has been delivered today.

Ordered on March 15th, received on March 27th morning - less than two weeks shipment (9 working days). Nice to point out: the shipment cost from China was included in the buying price ($198, this was roughly €195).

The package was surprisingly flat - despite a waterproof/shock-resistant packaging layer (thin, but good enough).
Here is the boxed device, once this shipment packaging is removed:




Inside the box:


As you may notice, the left bottom corner has been slightly bumped, but the thick inside foam did the job, and the protection is great on the whole periphery. The pieces are secured in the two small black bags with drawstrings (you may choose to use these for later storage, or store the pieces inside the device drawer).

Like the Centaur, the e-ink display welcomes you (despite the device being powered off):


Under the chessboard, the user's manual and a quick-start guide (isn't the logo charming?):


My first impression: the plastic casing, the board surface, the overall weight and rigidity, the e-ink display give a flavor of quality. Not high-end (it remains a plastic device) but definitely better than, say, a Millennium ChessGenius/CGPro.

The keys operate with a nice, accurate click-down sensation.

The sensitive board is very pleasant to use, very reactive. The colored LEDs (you may dim them, reduce them to minimal indications...) can coach a beginner and provide fun for children (they are attractive, and that's already an achievement).

The help this system can provide looks to be limited to the target squares a piece can reach (it won't tell you you should better move another piece - which is only true in training problems mode). I think it does not prevent learning, as for instance a red lighted target square only basically warns the piece can be captured, if moved there. It does not dive into checking there is an enabled re-capture, thus the move can be considered. But a beginner would therefore pay attention he can balance the red flag with something he plans.

The voice system offers uneven quality, to my opinion; with verbal often very neat and clear to understand, and sometimes scarcely recognizable. Anyway fun and adding interactivity for casual games. Language used here is English, voice is a woman's one. Funny enough, a bit of a Chinese accent that fits very well the device homeland!

The crystalline chessmen are stylish but won't fit everyone's taste. I like them, they comply with the spirit of this chess computer. They are a bit too light for best handling, but stand well on the board thanks to the powerful magnets underneath. And technically, as already stated about the board, the auto-response system is very successful.

I am still unable to share any clue about the playing strength, at least 15 out of 20 levels are designed for children and beginners, featuring instant response time.
Level 16 is close to 0.3-0.4 seconds per move; level 17 0.5 to 1s; level 18 2.5-5s (starting with this level, LEDs animation appear in the center four squares, until the move is computed); level 19 is around 8 to 16s per move; and level 20 (the strongest) looks to use roughly 45s to 1m30s per move. This is an average on 40 moves as an order of magnitude, including some instant moves from the opening library.

The clock system (optional, both in display and actual time control) appears to be independant from the level setting - meaning the computer won't adapt its thinking time to it, as far as I could check. It is therefore irrelevant to enforce a short time control with the highest level (20), it would unavoidably flag itself. It is nonetheless relevant with the so many levels that play instant (or so) moves.

Now I shall need time for any skills level assessment!
Cheers,
Tibono

[Tibono2]

Hi,
I can see there are some browsers suffering from a limitation preventing the display of the above snapshots.
As a workaround, a mouse right-click menu may enable opening the image in a new tab.
Otherwise I provide direct links here:
boxed device 1
boxed device 2
inside the box
powered-off display
manuals
Best,
Tibono

[samson]

Dear Tibono,

We truly appreciate your enthusiasm for both international chess and our product. Thank you once again. We have received your feedback in full and will work hard to correct our shortcomings and continue to innovate our product. Our goal is to create a comprehensive and better electronic chess game, serving the vast community of chess players and contributing our efforts to this field. Please continue to follow our progress. If we release any new products, we will offer them to you at a discounted price as a loyal customer. We hope to have your continued support. Thank you. If has any problem with L6, contact me any time. My email address is liaojunqu@vonsetchess.com

Best regards.
Samson

Hi,
I can see there are some browsers suffering from a limitation preventing the display of the above snapshots.
As a workaround, a mouse right-click menu may enable opening the image in a new tab.
Otherwise I provide direct links here:
boxed device 1
boxed device 2
inside the box
powered-off display
manuals
Best,
Tibono

[Tibono2]

Hello dear all,

I am back with some data and comments about the Vonset L6.

I am currently running computer games L6 versus other chess computers and chess programs, in order to normalize a CElo (Computer Elo). As my standard for timing is 15s per move on average, the L6 plays at level 19 (of course, the opponents as well are set to an equivalent pace). This will need several more weeks to complete.

In the meantime, I also ran the Khmelnitsky test with the L6. Yes, the L6 provides the necessary features that enabled the test!

Basic requirements are a position setup ability, and a score display. The score is displayed in centipawns, which is accurate enough for the test; much better than some chess computers displaying their score in pawn units with only one decimal. The process for setting a position is smart and very efficient, this helped a lot as I had to setup positions (from scratch) more than 300 times... I know what I am writting about!

You may wonder why the 100 positions, 200 questions of the test triggered more than 300 setup process? This is due to a misleading behavior (I consider it a bug). On setup completion, you define the color to play next move, and the color for the 'bot' (=the L6). If the opponent plays from the position that has been setup, the L6 uses a quick evaluation in order to assign and display a score for the player's move; then the L6 computes its answer correctly, leveraging the playing level it is set to. The other way around, if the L6 plays first from the position that has been setup, it plays too fast and provides a doubtful score, maybe the quick evaluation one. Next moves are evaluated correctly. I had already performed 70% of the test when I realized this score was sometimes weird, did dedicated tests, and got evidence of the issue. The workaround is rather simple: just set the position one move ahead of the desired one, always take the first move as the player, perform the move reaching the desired position, and the L6 thinks at full power. This looks cumbersome, but don't forget: the L6 is designed for playing games, not for analysing positions. Another limitation in the setup mode: it can only start from scratch = an empty board. Not a blocker, you can always rebuild the position you want (and the process is fast and efficient, as already stated); but a choice between empty or current position, as a starting point, is a "nice to have". And enable/disable castling would help with some positions (it is disabled in the current software).

About full power: I use 3 minutes thinking time as a standard search time for the Khmelnitsky test with computers; so I ran the test with level 20 (the highest) for the L6. It is both fair as it is the best the L6 can do, and unfair for comparison' sake with other computers, as the level 20 uses much less time than 3 minutes for computing.

Now to some data: I include pictures within the text (my Firefox browser displays those), but I also add links to enable access for other browsers (It can be a security limitation, not to embed pictures from another website, and a non-SSL one in addition).

The Vonset L6 reached the 1441 score-mark on the Khmelnitsky test. How does this compare to well known, classic chess computers? Well, it is the very same score as the ones of the Fidelity Excellence 3Mhz, and the Saitek Kasparov Blitz (despite the reduced thinking time used by the L6). On the current Wiki-Elo-Liste of the www.schach-computer.info website, these are respectively ranked 1795 and 1771 CElo (at active chess conditions).

As I ran the Khmelnitsky test for 29 chess computers so far (not including the L6), I could compute a calibration function from the below graph:

link: Calibration

From this graph, 1441 KT provides a forecasted CElo 1774.

More reliable than the overall evaluation, the distribution of chess skills is:

link: KT_L6

The skills that stand out most are: Threats recognition and Calculations; then Tactics is also a strength. The overall impression of fast computing (the device is very reactive, and able to play very fast) is confirmed: the embedded CPU must be somewhat powerful.
As usual for computers, the L6 is more comfortable with counterattacks than with attacks. Strategy is a bit weak. A lack of knowledge obviously impacts standard endgame positions, and endgame more globally. Nevertheless, the L6 performs better than the Blitz in these endgame domains!

Here is a comparison graph:

link: Computers comparison

And a comparison to a same level human player:

link: KT_L6_Player

Well, despite better attacking skills, the human player shoud play cautiously (beware the counterattacks, if any tactical concern is missed!) until reaching the endgame, where he should prevail.

A last but not least observation: the battery capacity is large enough for hours of testing activity!

Up today, I am very pleased with the L6. It is really nice for casual games; is not designed for analysis but nevertheless can "pass" the tough Khmelnitsky test and reach a KT-score that competes with the Excellence and Blitz ones, despite its fast response time. From my collection, there are now two computers that are attractive enough to suit being played by my grandchildren: the Lexibook Chesslight, and the Vonset L6. But the L6 is definitely better on all main aspects: playing level, moves registration, pieces stability, coaching lights and overall attractiveness for children, beginners and casual players.

I'll be back with a more accurate CElo (Computer Elo) evaluation once enough games are completed. Expect it to be lower than the above forecast, as games are played one level lower (19), to comply whit the time limit I use for computer chess tournaments.
Cheers!
Tibono

[Fernando]

Reading your examination, Tibomno, I concluded that it is not the one I want. If I want to play an Excellence kind of computer I will get one on eBay.
Thanks!!
Fern

[Tibono2]

All,
I am still in the process of tournaments for Computer Elo evaluation for the L6, set to level 19.

On another hand, the adverts for this product obviously target parents keen on offering a chess opponent and coach to their children, and the design itself complies with such a use. As Samson wrote in this thread, I quote: 'Our brand concept is to apply the power of technology, especially artificial intelligence, in the field of family companionship, to accompany children and the elderly.' And indeed, I plan to leverage the L6 to raise some interest about chess for my own grandchildren.

That's why I wanted to also test the lowest levels in some way. Sparring the L6 with my Lexibook Chesslight has been a nobrainer to me!
I started from the lowest available beginner level from both devices, level 1 for the Vonset L6, level A1 for the Lexibook Chesslight (they range from A1 to A5).

My plan was to increase the Vonset L6 level after each game, whilst increasing the Lexibook Chesslight level only after a win by the Vonset. The Vonset always played first (white color, so).

Vonset L6 levels 1 to 5 were dominated by Lexibook Chesslight level A1 (5 losses for the L6).
From lv. 6 to 10, Vonset L6 mostly dominated the Chesslight A1, but did not win a single game (4 draws, 1 loss). It just avoided any win, either reaching stalemate or three-fold repetition. Only level 11 for the Vonset enabled achieving a win after its large domination over the Chesslight A1.

This gave no clue about the respective levels, I could just notice the Vonset play was indeed, step by step and level after level, getting slightly "stronger" (ok, rather "less weak"). So I decided to change the process, and raise the level for the Lexibook not only if it lost a game, but also in the advent of a draw. This means a draw would be considered a success for the Vonset, and trigger an increased challenge for next game.

I thus resumed the games after the first draw, achieved by level 6 vs A1, with the game Vonset lv.7 vs Chesslight lv. A2.

The Vonset was dominated, and lost. Next two games were losses as well, Vonset L6 lv. 8 and 9 dominated the Chesslight A2, but standing by the ostensible commitment not to win, finally ended as losers. L6 level 10 dominated as well, up to three Queens circling around the opponent King alone, carefully avoiding the mate, up to 3-rep draw.

The above draw let me increase both levels, L6 to lv.11, and Chesslight to lv.A3. The L6 dominated, but blundered and lost to a corridor mate. Lv.12 struggled to balance the game, then dominated, and again lost to a corridor mate.

Level 13 won convincingly (still against level beginner A3), so level 14 had to play Chesslight A4: the game was balanced until middle game, then dominated by the Vonset, and won again. Therefore I kept the level 14 unchanged for the L6, and raised the Chesslight to level A5, but no way: the Vonset easily won again.

As a conclusion: levels 1 to 10 can be named "play and win", they are just sparring partners that won't hurt you, as a golden rule. Maybe an interesting starting point for children, assuming you want to take care of their motivation.
Level 11 is the milestone for a first bit of a challenge: the computer will be kind, will sometimes blunder, but will achieve a win if you let it excessively dominate. To my humble opinion, a must for learning purpose, and for real fun. The playing level then raises noticeably with each increase in the level setting, despite instant response until level 15. I would therefore consider levels 11 to 15 as the truly interesting training levels.

(Of course, I can make the games available as a pgn file, should anyone want them. Just PM me.)

Enjoy a nice Easter weekend,
Eric