Go9dan.com

[oren]

(I have a feeling NASA's Curiosity has followed a similar approach: baby steps, test test test; more baby steps, test test test; repeat.)

Off topic, but not really. Systems level testing of Curiosity was virtually impossible, so the system level tests were all modeled. Piece components did have a lot of testing done.

[macelee]

It does look like they spent too much money on marketing and too little on the technical side. Even now some pages on the site are down, displaying error stack trace, therefore revealing the technology they are using, and showing a simple programming error to the whole internet. The word amateurish has been used in the thread. This is a very basic mistake and a big security concern. For a game server the security of the system has to a matter of paramount importance.

[LocoRon]

it is not fair to compare Kaya's early technical issues with go9dan.com's (Kaya's problems seem to be...well there are already threads about that). Kaya's development model leans heavily on user testing. This was made clear upfront and they have visible features on their site that actively seek feedback.

Sorry, but I disagree. Comparing the technical problems of these two sites is far more fair than comparing it to, oh say, Google or Twitter... but hey, they both have their issues as well. Ever been unable to check your gmail (through the web interface) or caught a fail-whale? It happens. Things mess up.

And if you expected go9dan to have no (or only minor) problems, well, that's kind of your fault. It was announced as being a beta, and the website's banner still carries the beta tag. (Although, I will grant that, from what I've read of it, it seems more like an alpha right now than a beta.)

[billywoods]

Sorry, but I disagree. Comparing the technical problems of these two sites is far more fair than comparing it to, oh say, Google or Twitter...

Perhaps lemmata's point was simply that Kaya did not expect big things from day 1, and went through extensive beta testing before being publicly released even by an inch. While go9dan claims to be in beta, it is acting like it's the biggest server in the world. It's not - it's a botch job. It seems to have no eye on usability, security, decent features or the like. The only thing it has going for it is this exciting series of matches, but so far it's drawn a big crowd and disappointed them twice. It doesn't look like an alpha or a beta to me - those words imply that it is being alpha- or beta-tested - I would say that it's just crap.

Personally, I think it's a bad money making scheme. I don't trust it in the slightest yet, and a series of games won't change that.

[lemmata]

Sorry, but I disagree. Comparing the technical problems of these two sites is far more fair than comparing it to, oh say, Google or Twitter... but hey, they both have their issues as well. Ever been unable to check your gmail (through the web interface) or caught a fail-whale? It happens. Things mess up.

And if you expected go9dan to have no (or only minor) problems, well, that's kind of your fault. It was announced as being a beta, and the website's banner still carries the beta tag. (Although, I will grant that, from what I've read of it, it seems more like an alpha right now than a beta.)

I suppose you and I disagree on the degree of the problems on go9dan.com. I would not characterize these problems as minor. When a server with a small user base is unable to get a pre-arranged and pre-announced (by the server!) game going between two pros for more than a 40 moves without technical issues interrupting the game (likely the only game on the server at that time), then that is a major problem in my book.

I did not expect go9dan.com to have no problems or even only minor problems. I did expect to be able to see the game between Lee Sedol and Andy Liu that they told me that I was going to be able to see. I did not expect to be able to play a game myself or chat. Forget about me watching the game, they couldn't even get the two players to complete a game.

Another major difference between Kaya and go9dan.com is context. While both carry the beta tag, the latter announced that it was going to run an online league between ten top pros in China and Korea while the other said that "they're working on things". The beta tag itself is meaningless. There are ways to implicitly communicate the readiness of a business operation to the public. Go9dan.com chose to project the image of readiness by announcing ambitious projects that require readiness for the extremely near future.

Of course, I don't expect Go9dan.com to be perfect from the beginning of the development process. However, the problem is that they raised expectations.

The online pro league is supposed to have total prizes worth $100,000. Despite the promotional nature of the league, it is serious business. The league is supposed to start this week. If Chen Yaoye beats Lee Sedol by time because of lag and a dispute arises between the players, how will it be resolved? The winner of the Chinese Mingren tournament receives about $4000. By go world standards, this is not a small event. What if the problems occur multiple times? Does it delegitimize the event? Will there be ill feelings? An international incident? Will anyone drop out of the league? How many times will pros tolerate having to restart their games from the beginning? Will some not play seriously if this becomes too much of a problem?

Although both sites are "in beta", the scope of readiness that is needed for go9dan.com's announced plans for January is much greater than the scope readiness that is needed for Kaya.gs's announced plans for January.

None of these things really matter to me personally. I am not even angry at go9dan.com. My hopes were high, but my expectations were low. I am just disappointed that people with good ideas, capital, and connections seem to be squandering a golden opportunity by running the business and technical operations as if they were complete amateurs. Perhaps they will get their act together before the go league starts. Who knows. I hope so. [edit:sp]

[LocoRon]

it is not fair to compare Kaya's early technical issues with go9dan.com's (Kaya's problems seem to be...well there are already threads about that). Kaya's development model leans heavily on user testing. This was made clear upfront and they have visible features on their site that actively seek feedback.

Sorry, but I disagree. Comparing the technical problems of these two sites is far more fair than comparing it to, oh say, Google or Twitter... but hey, they both have their issues as well. Ever been unable to check your gmail (through the web interface) or caught a fail-whale? It happens. Things mess up.

And if you expected go9dan to have no (or only minor) problems, well, that's kind of your fault. It was announced as being a beta, and the website's banner still carries the beta tag. (Although, I will grant that, from what I've read of it, it seems more like an alpha right now than a beta.)

I suppose you and I disagree on the degree of the problems on go9dan.com.

Probably not actually; hence my saying it's more like an alpha than a beta (in other words, very far from finished).

The only thing we might disagree on is whether or not it is fair to compare go9dan's technical problems to kaya's technical problems. I think it is fair.

Personally, I think [go9dan]'s a bad money making scheme. I don't trust it in the slightest yet, and a series of games won't change that.

Agreed.

[Javaness2]

What's wrong with having a commercially successful Go server? I think Go9dan launched too soon, maybe it will recover given time.

The best html based server I have seen is still Alex Selby's.

[mlund]

I wonder if this was primarily a load-handling issue. When things look fine internally and then implode once you bring in a large audience it's usually a load-handling issue.

Goodness knows if they are using a pure Client-Pull method via AJAX or a steam to update all the viewers this thing is liable to implode on itself. You need the viewer architecture to load on the client side once and something like COMET to push out every new move or comment to the audience. Ideally it's just a 4-character data point for the move and a string for a comment. Also, you'd really like to have the push to the audience running on a separate server with separate bandwidth: ideally one set up for running the game (with the security, I/O, rules logic, and archival logic) and another set up for webcasting game data once it hits the archive.

Streams and Data Pulls and failure to compartmentalize your game processing vs. broadcasting work-loads seems like a recipe for disaster.

- Marty Lund

[billywoods]

What's wrong with having a commercially successful Go server?

Nothing. But:

• I hugely disagree with the idea of anything that is commercially successful without being fit for purpose.
• I disagree somewhat with businesses that crowbar themselves into a gap in the market which doesn't exist. There are free alternatives to go9dan, but go9dan has capital and big names behind it, is advertising itself hugely via this series, and will be hoping to gain momentum and make itself known as the place to play go. Why else would anyone play there? It has nothing that KGS doesn't have, except for a fantastic ability to advertise itself. I'm confident that it won't be the death of KGS in this case, but this kind of thing isn't unheard of.
• I disagree with a business that puts money before its customers. Look at their website - I know how much I should pay for what features, and I know which price option is most popular(!), but I can't even play a game. Can you honestly tell me you trust them to value you as a customer?

In short, I view a commercially successful go server as a go server (run for the sake of go) that happens to be commercially successful. This is a few people who wanted to start up an enterprise (run for the sake of money) and happen to have connections in the go world. Well, that's my rather cynical take on it, anyway. We'll see if that changes.

[badukJr]

They forgot to remove the misspell mark from their promotional pictures

[Xa17u]

The best html based server I have seen is still Alex Selby's.

I don't know about it, care to explain?

[daal]

Lee Sedol vs. Chen Yaoye started about 20 mins ago...

[billywoods]

I get a badly spelt error message.

[daal]

Works for me. Are you registered? To find the game, you have to click "play now" which brings you to the lobby where the game is being played. Double click opens it.

[billywoods]

Works for me. Are you registered? To find the game, you have to click "play now"

With you up to this point - and here I get:

"Error
You are unable to connect to 'Lobby Server' at this time. Please refresh your browser. If it doesn't help the issue, please wait for server is recovered. Sorry for inconvinience.CLOSE"

Needless to say, refreshing does nothing.

No matter - I wouldn't have stuck around to watch it myself - but my (cynical) guess is that the problem of heavy traffic flow has been patched over by setting a cap on the number of users who can join, and once the lobby is full any new people trying to join are denied connection automatically. Shame.