Interview with Mike O'Brien and Pat Wyatt

I agree that that it sounds odd, but they really did say there was a single BLOB per character, plus a further single BLOB per account. While I question your calculation that data is as small as 120K per player, you're surely not off by a huge factor. And most people who'd be over that would probably be the kind of serious player who owns multiple chapters, and it's not clear that more than a few million copies of all chapters COMBINED have ever been sold.

Bottom line on that: It sure would seem that the core player/account data is in the low hundreds of gigabytes at most. And that's consistent with the sense I got from them that the amount of DBMS scale-out hasn't been that significant to date. (Note that one might scale out for reasons of throughput long before one maxes out disk capacity; on the other hand, they clearly prefer not to pay extra DBMS license fees if they don't have to.)

Also, please note that at several thousand players/server and 2.5 gigs RAM total per server, the BLOBs really couldn't be a lot bigger than the figure I quoted.

Anyhow, it was a rushed call, and I asked for followup. After being ignored on that request for a week or so, I decided to go ahead and post based on what I had.

Arggh. I answered the size thing, but the note seems to have gotten eaten somehow.

EDIT: Double Arggh. Now it IS up. I'll just leave both posts up now, despite the partial redundancy, in the hope that helps with clarity.

----------------------------------------------------------------------------------

REDUNDANT PART:

Yes, I do mean one BLOB per character, plus a second per account. That limit is consistent with the number of players per server, where RAM is constrained. Yes, no matter how you do the arithmetic it is hard to see how the total database size could be more than a few hundred gigabytes tops. But then, I also got the sense they used a very small number of servers for the core database.

NEW PART:

As for auction house possibilities, the most detailed thing I wrote is All of the above is spelled out at http://www.dbms2.com/2007/06/09/the-...of-guild-wars/

I think that this speaks volumes -- in multiple ways.

It really goes to show just how good these guys are to build such a great game while learning the tools as they go.

Unfortunately, the lack of knowledge looks to have also resulted in some poor database design decisions. Without knowing exactly how the tables and "BLOBS" are actually designed, it's difficult to be sure what the problem with an auction house really is. Even the interviewer (OP?) admits not clearly understanding the issue.

But it would be really nice for the Anet development team to simply own up and say something like the following:

"We learned a lot developing Guild Wars, and have done a lot of things right. However, certain incorrect design decisions early in the project have made the creation of an auction house a project that would not be a fiscally responsible expenditure. Rest assured, however, that the ability to auction goods is a core design consideration in Guild Wars 2 and this feature will be available in the final product."

Or, more simply:

"We screwed up and now it would cost too much [payroll] to implement an auction house in GW. We'll get it right for GW2."

How about just being forthcoming about it Anet?

-Forjo

[QUOTE=Forjo
How about just being forthcoming about it Anet?

[/QUOTE]

I think that A.Net might have ben a lot more open with the community and able to say "stronger" things if they were not a wholly-owned subsidiary.

Anyway, this was a neat article. Thanks!

1. I agree with your favorable comments about the developers, and your reasons for them.
2. Yes, I'm both the OP and the interviewer/author behind the posts linked.
3. I think they have been forthcoming.

They've used the word "possibility" far too much to be considered forthcoming, in my opinion. After all, isn't Gaile's response to most questions about features, "No plans for it, but it's a possibility"?

How about they just say, "No plans for it, and we're probably not going to do it. Don't hold your breath." They did it with this interview, and that's a step in the right direction, but really, they've been too "uncertain" with us.

Great articles though, they explained a lot. I was particularly annoyed how the developers didn't know SQL when they started making the game...no wonder the game seems to be getting worse to many people. They didn't have the knowledge necessary to support such a project when they began, so that once things got complicated, now they're having problems. With all the skills, maps, and features that have been unveiled since relase, it's easy to see why they haven't been keeping up like they used to thanks to this interview.

In my opinion, they shouldn't have released all those features, skills, etc. unless they could support them as much as the original game, but let's not turn this into another topic about Anet and incompetence. I'm just venting here.

I admit to not being a close watcher of Gaile's postings, but I think her comments were in line with a reality of "Well, we considered it, but upon closer examination of what was involved decided it wasn't going to make the cut any time soon."

As for not knowing SQL (but having built a DBMS once, something I didn't mention in the posts) -- I bet that's very common for a game developer.

Finally, suppose they built an auction house -- the question isn't just the back-end. This isn't a very interesting idea unless there are lots of items up for auction, right? Well, what do you imagine the UI being like for that? I imagine it's a much more detailed transactional/inventory UI than anything else that's even been seen in Guild Wars. Well, how do you imagine that would be integrated into the rest of the UI?

I'm not sure they could do the auction house UI with off-the-shelf technology, whether application or dev tool. And if they have to roll their own -- well, again, that's a whole new level of complexity in transactional technology.

And finally -- it is VERY common for highly smart and able programmers to not understand the difficulties of a transactional system until they sit down and actually try to design it.

Thanks for the info, Francis.

"No Auction houses in GW 1 because it's unrealistically complicated to implement in to the game."

*Trolls march in*


As previously mentioned, everyone here would really, REALLY appreciate these kind of direct interviews over vague, general statements. It's alienating to players when Anet reverbrates cookie cutter phrases or avoids subjects. This just opens things up for critics and trolls and makes loyal fans and other players feel betrayed.


Not only would admitting fault and a screw-up make people feel less alienated and betrayed by Anet, but it would serve as an image builder as well. Planning an auction house for GW 2 and promising it, as well as keeping that promise, would be a morale boost to those who've lost faith in ArenaNet and hope in Guild Wars.

Nice read. Thanks again Francis.

The 'incorrect' design decisions you imagine are likely the same ones that allowed them to build a system that supports a truly global character database, making GvG between players anywhere in the world possible. I doubt they see those decisions as being incorrect.

I see it even more as being a matter of priorities.

They solved or averted a lot of problems by being super-simple-minded in their relational database usage and design (not all of which I spelled out because, among other reasons, I didn't get all the details). The negative side effects are, in their expert opinion, a small price to pay for the benefits.

Remember, folks, no matter the BLOB size, all those data still have to get moved around, and with that comes loads of complications along the way. We're not just talking staionary data, but rather dynamic transition from one server to another, etc.

And lastly, before I forget again: A welcome read, Francis, very well done! Thank you for sharing that with the community!

Francis, thanks for the insights. It confirmed a lot of what I'd suspected made up the technical side of GW.

I think before the trolls flame Anet for not being more straightforward about the reasons behind the auction house, let me add this perspective:

Anet knows that it's not technically nor financially feasable ATM to make an auction house. The community at large is demanding one, and so they've got to make a statement of some sort that'll please the masses while simultaneously giving their competition and detractors 'live ammo' as it were to use against them .

If Anet comes out and says that, then WoW, LOTR, and all those companies and games could potentially use that statement in advertisements, promotional material, or to potential sponsors of contests and such. It even gives people like forum trolls something concrete for them to latch onto and bash them with. Furthermore, if Anet was able to get the resources later on after that statement was released and got an auction house, it could be too little, too late when dealing with potential players and an already miffed player base as it is. This could even cause people to start going through a list of things that Anet also said couldn't/wouldn't happen and start asking 'Well, you said the Auction House would never happen, but it did, so why not introduce X?'

It sounds to me like it's not impossible for them to make an auction house, but one they don't have the time and resources to devote to it at this moment. Besides, GW has survived two years in release without one and is continuing to expand. An auction house would be nice, but the lack of one obviously hasn't killed them either.

Speaking as a player who enjoys farming a lot -- folks, please get real. What's the effect of not having an auction house? Basically, it means there's no great way to cash in on your loot.

But what would an auction house really add? Would it provide a way for you to make more money and hence buy more stuff you wanted? Don't be so sure. If everybody could sell more efficiently, the various effects on the economy and on your wealth would likely counteract each other to a large extent. Right now, anybody who's intense about farming can buy almost anything they want, but not necessarily the top-end super-rare items. If there were an auction house, exactly the same thing would probably be true.

And if you're looking to an auction house to make make the overall economy more efficient and hence make having-the-most-money be an accurate barometer of a kind of game success -- uh, do you really think that? There are all sorts of other reasons why that will never happen. Game wealth as a measure of success is a VERY imprecise metric, and that fact wouldn't change if gBay were built as the most sophisticated auction tool on the whole Internet.

Yeah, I wish I could easily liquidate the four Darksongs I have cluttering up inventory -- but if I could, then so could everybody else, and with that much supply the price would be next to nothing anyway.

You -- and everybody else -- are welcome!

It was a lot of fun.

That's not at all why I would want an auction house (though I'm fine with any type of good trade improvement). The very basic reason is not pursuit of money, it's pursuit of more enjoyment out of the game. In other words, how trading is performed now, standing around a "crowded" city and "yelling" over and over what you have for sale until just the right person walks by that actually wants that same thing your selling, is severely not fun. Doesn't matter if you undercut the item's worth tenfold, if there isn't a buyer, you won't be able to sell it, and henceforth waste a lot of time you could be spending, ya know, having fun - in a video game.

So your options are:

1) Waste your time doing this;
2) Use a third party website (like guru auctions), which tends to take a lot of time as well and can be too clunky(setting up the advertisement, screenshots, setting up meeting times, don't do it just right and get temp banned); or
3) Just keep having fun and sell everything you don't use to the merchant. Which of course means you'll often be short on cash to enjoy some of the more pretty things (one of the main reasons people PvE), like 15k+ armor, dyes, keys, etc.

With an improvement on trade (like a simple player shop, or ability to place more items and more characters in the party search), you can add an option 4 - set it and forget it. Would be like AFK title-farmers. Sell while you're asleep, or doing other things, ya know, like school and work.

People have said over and over again it doesn't have to be an auction house. For some reason, most other suggestions, good ones at that, keep getting buried by the mass for and against auction houses. I often wonder, is Anet aware that there are other options that wouldn't include a mass re-working of databases or whatever? Surely they can think at the edge or just outside the box? A little creative thinking can go a long way, folks.

Utterly standard items -- greens, mods, inscriptions, even minipets -- could probably be handled on the current trader system, I would think.

Truly unique items -- golds -- rather than "unique" items -- greens -- wouldn't be practical to handle that way, I think.

Hell, honestly, more traders for mods, inscriptions, maybe even greens would be nice. As far as player shops go, suggestions have been made by others and myself as well, to have a separate window like the party search (or even within the party search), that has links to the player's shops in that town, or even a central trading outpost. When others visit your shop, there maybe is room for 5-10 (or more) items you can list with prices. If someone wants to buy something, they just message you and you trade as usual.

Do you think, given what you've learned about their system, this would be easy to implement? Would it require a lot of server power if this was in place?

Never mind the back end for a moment; I'm concerned about the UI. Suppose the shops were used by a lot of people in the same town; browsing could be a real pain. Or if they weren't used by a lot of people, what's the big benefit of implementing them?

As for the back end -- first of all, if all you're doing is giving a way for players to add messages to a town environment, there is no true back end issue. That said, a couple of different possible approaches come to mind. One is longer text strings; the other is actually putting items into something like the current party window. The latter would, beyond UI, cause more development work than the former, I think. But the former might require well over 1K/seller to satisfy people, which could be a lot of stuff to push across the wire if there were a bunch of sellers. There also would be a major UI issue.

I just think that any way to better support the trading of arbitrary items would require significant development effort.

Well, that article finally explains why things like hairdressers and updating existing items are such a big deal for them. I had always assumed it was just a case of Scotty-Sydnrome ("I canna giver any more captn") rather than actual technical difficulties, but if your database is effictively just a transactional file system, then yes, I could imagine how doing such things would be difficult.

Coming from a developer in the banking sector with zero game design experience, it does seem like a very "odd" design choice, but for the team that developed such an awesome game, I'll overlook it

Well, let me illustrate what I'm thinking, and you tell me if this seems like a huge problem for either the UI or the back end:


When advertising in the already existing party search window, nothing much changes here except the little button at the bottom that allows people to open up a separate window, the "player shop" or whatever else it can be called. There may be other things added, I added a little counter in the popup that says how many items are listed.


When they click the button, they are brought here, as a separate window. It simply lists what they player wants to list, and they can whisper the person directly from the list, or close the window - that's it. There could be more functionality added, such as "drag-and-drop" items into the list and have them be automatically added. If this is implemented, then simply scrolling over the item can give a popup similar to what would normally popup if scrolled over in the inventory.


Editing the items would be a snap, popping up a small window with editable text, like so. Of course this would only appear for the seller. If the "drag-and-drop" is implemented, this window would be even simpler - only "delete" and "close" buttons would be needed.

Simple. Elegant. Useful as hell. I would kiss the ground everyone at Anet walks on if they could just do something like this. What do you think?