Hactar 1.3 versions

[lpaatero]

Hactar 1.3 introduced stronger engines as subscription options.

Engines are possible to try for at least 7 days for free.

Strong engine is about 1kyu at 19x19 board.

[lpaatero]

Notes about 1.3.5 version:

Server-based capture go is available as part of "go master" subscription.
Level of server based capture go engine is not as strong as I would like. It does not feel much stronger that phone based.

Now server calculation is using 50 times more calculation compared to maximum level in phone. It is quite possible that additional heuristics are needed to make engine stronger, even with more computing power.

I cannot promise when and how much I will have opportunities to improve engine. Improvement work is quite time-consuming...

Also note that this is still beta-quality; please report any problems!

- Lauri

[noorudin]

I have some suggestions for the Problem mode in Hactar, as following

1. for some kids like competition, is it possible to build a scoring system in the Problem mode of Hactar? And the scoring record could be sent to the ranking list in the server. Someone who have subscribed the problem collections in Hactar could compete each other.　For example, in the scoring system, the stone position which was far from the right position the less in score...like that.
2. Is it possible to create resolving options (black play , white play and play as both) in the Problem mode of Hactar.

[lpaatero]

Hi,

sorry for slow reaction. I missed this post due to other things happening at the same time.

This tsumego competition mode could make sense; in particular if there would be new problem sets coming out.
Have to think this bit more...

Right now it is not possible to select turn etc for problem.
Would you like to have possibility to start from same problem as black and as white?
Or does Hactar guess the start color wrong, and you would like to correct it?

Lauri

[noorudin]

Right now it is not possible to select turn etc for problem.
Would you like to have possibility to start from same problem as black and as white?
Or does Hactar guess the start color wrong, and you would like to correct it?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
the HACTAR is very good! the HACTAR is correct!
For tseumgo mode, could I have some options to check [ ] black to play [ ] white to play [X] alternation by myself<----(ex. "B-W-B-W-.....-end",played by my self, if finishing the game, the HACTAR give me a score, means "guess both")?

Why did I need the "alternation by myself" function? For somebody reviewed the game "XXXXXXXXXXXXX.sgf" by his memory . It can help him by this evaluation function.

[lpaatero]

So instead of hactar playing other color, one could play both;
or alternatively hactar could try to solve the problem and user prevent it?

Sounds sensible. There seems to be unfortunate problem; current tsumegos do not contain enough variations for these to work nicely.
I think machine-generated variations would be needed.


About tsumego competition:
I am unsure what it should look like (to user). These kind of things are bit unfamiliar to me.
Would it make sense to compete for getting "rank"?
eg Hactar would "estimate" rank when solving. This would at least give some goal to tsumego.
Same rank could be applied to playing also.

Lauri

[noorudin]

So instead of hactar playing other color, one could play both;
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Sure for this. This is what I mean in the last of three options! It's correct!! (or like this pattern:[ ]random [ ]guess both. The user could check one of them only)


or alternatively hactar could try to solve the problem and user prevent it?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
No, this is not my thinking.


Sounds sensible. There seems to be unfortunate problem; current tsumegos do not contain enough variations for these to work nicely.
I think machine-generated variations would be needed.


About tsumego competition:
I am unsure what it should look like (to user). These kind of things are bit unfamiliar to me.
Would it make sense to compete for getting "rank"?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
It may look like "% of completion of a problem" , but the format of Score is just the digits of percentage(except "%" symbol) , finally get the summation of the collection.

In Tseumego mode, somebody could build a tseumego collection (called "a Tesumego room") and uploading his own Tesumego collection to the Hactar server. He/She invite his friends( sometime they are getting the similary rank and strength in the real world) get into this room to resolve and comepeting, just for fun. The Hactar collect these tseumego collection become a growing Tseumego Database. The correction of a problem was provided by Feedbacks Statistic. The Hactar can offer the man who uploading a Tseumego collection some reward.


eg Hactar would "estimate" rank when solving. This would at least give some goal to tsumego.
Same rank could be applied to playing also.

Lauri

[lpaatero]

The tsumego idea sounds good, thanks for it!
Making it requires quite a bit of update to current server side, this probably delays doing it somewhat.
I am currently cleaning up server side anyway, so after this I try to take a look at this.

regards
Lauri