Life In 19x19

Bill Spight wrote:The fact that sometimes averages give terrible results is a big clue. Maybe averaging is not a good idea. Even when it appears to be.

spook wrote:Merging statistics of 2 different bots is as difficult as making 2 dictators agree on something.
Perhaps we need to train an AI for it. :scratch:

OK, so a simple average isn't quite the right thing. But don't give up! Combining different models can often outperform the individual models.

spook wrote:

Jæja wrote:I see that bot analyses and variations are stored within saved games. Do you plan on allowing for these to be made public, e.g. by sharing a URL, like a Google Drive document? This would be absolutely amazing

Not only do I think this is a great idea. I think it would help promote zbaduk a bit as well.

It would then also make sense to allow guests (i.e. visitors without accounts) to use the review tools. (perhaps in a read-only mode)

It's in my top prio list.

Sooooo cool. Thanks again for all your hard work and good luck with the project!

I've finally figured out that it is possible to use "stop analyzer" to show the server that you're "just idle" rather than "completely idle". I guess I had simply assumed that stopping the analyzer would reset the analysis like the ownership tool does? Very happy that this is not actually the case. More of a "pause analyzer"

Unrelatedly, sometimes (maybe 1 out of 20?) the analysis will be running merrily, but then I click its favourite move onto the board and that will cause it to disconnect, showing no statistics chart for 10 - 15 seconds before starting analysis of the new move at 0.

EDIT: As of a few hours ago, ZBaduk does not display "winrate" in the chart anymore, or only "decision"...?

EDIT 2: Nevermind, this just means you've already made it customizable! Awesome, thanks!!

I just installed an update.
Enjoy !
(PS: @Maharani, this should also fix the confusing gray chart issue.)

Maharani wrote: EDIT: As of a few hours ago, ZBaduk does not display "winrate" in the chart anymore, or only "decision"...?
EDIT 2: Nevermind, this just means you've already made it customizable! Awesome, thanks!!

Exactly, that was actually part of a previous release (was it last weekend ? )

One detail about the column selection: the selection of columns is stored in your webbrowser, not on your account. The reason for this, is that you can select a different set of columns for your mobile phone or tablet. But it also means that if you use both chrome and firefox on your computer, you may have to configure it twice.

---

(As for the game properties window, the selection of ruleset is displayed, but intentionally disabled.)

Another wonderful release. You're really pumping out those features like crazy!

y501 wrote:Also, do you still need someone to translate your site to Korean?

@y501 , sorry for my delayed response. Yes, I could use some help with Korean.
Actually, I have been struggling with translations for almost half a year now.
And one part of the problem, is that the website constantly changes.

I actually think that translations could be a lot better if I would allow (certain) users to directly edit text messages on the website.
It basically means that you can just right-click on any text inside the application, and then a pop-up would appear to enter a better translation.
It would roughly look like this:
I will post some more info about this in the following weeks.
And yes, if you are a native Korean speaker, you could certainly help me out here.

Thank you for your support everybody !
Take care !

Hello,
First of all thank you for adding Katago with variable komi. I can study my korean yearbook with the right komi. I have a question regarding that.
When I choose katago with 6.5 komi there is a difference between the chinese rule (black is around 53) and the japanese or korean rule (black is around 49), is it normal ?

Yes, this is expected. With 6.5 Chinese, consider the games where black gets 183 points on the board and where white gets 177 points of the board. (183 + 177 = 361). Black wins by 0.5.

In some of them, white plays the last move, so black and white will have played an equal number of moves, so the Japanese score will be equal to the Chinese score, Black wins by 0.5.

In some of them, black plays the last move, so black will play one more move than white, so the Japanese score will be one point less for black, White wins by 0.5.

It is possible that there is more than 1 move difference, or that white plays more moves than black, or that the score sums to an even number instead of odd number, but these are all less common. So mostly with this specific komi the Japanese rules will turn some Chinese wins for black into wins for white, decreasing black's chances.

thank you very much, very clear explanation ^^

Something very strange is going on with ZBaduk. Using KataGo as the analysis engine on a game with 6 stones handicap, white is given 100% winning percentage, instead of black.

Jæja wrote:Something very strange is going on with ZBaduk. Using KataGo as the analysis engine on a game with 6 stones handicap, white is given 100% winning percentage, instead of black.

Screenshot 2020-03-28 at 12.35.04.png

Does it change accordingly with Black to move?

Strange, I am trying to reconstruct, but without succes so far.

With just 16 playouts that is of course really low.

Is this a screenshot of today ?
Earlier today (March 28th) there was a problem with our load balancer, which slowed our system down.

Are you able to reconstruct this issue, or was it a one-time occurrence ?

Hello, have some trouble to use katago. Am I the only one? Just wanna analyze an even game with 6.5 komi.

lichigo wrote:Hello, have some trouble to use katago. Am I the only one? Just wanna analyze an even game with 6.5 komi.

Should be working fine right now.

(PS: I prolonged all active memberships with 3 days to compensate for the hickups.)