In depth look at Auction House feasibility

Okay, for one, selling 3,000,000 copies does not mean that all 3,000,000 players are on at once. Moreoever, a lot of these copies are also probably linked accounts, with up to 3 copies constituting a single account.
Two, text compression is amazing. The 3MB file that OP quoted could easily be reduced by several orders of magnitude. Also, a single number does not take an entire byte. If you do a little math, a single number should take up no more than log(base 2)(that number + 1) bits rounded up (note that 8 bits make a byte). (The 1 is there as a special null value). With actual compressing, that 3MB file could turn into a mere kilobyte or two because a lot of the mods will be the same. On the technical side, there is nothing that would be unfeasible to implement an auction house.
On the other hand, concerns with intellectual property from previous developers of blizzard are the larger matter. I doubt that anet wants to get their ass sued.
In general, load issues and even search issues aren't a very large matter. Technical issues could easily be solved. A testament to this is just how guild war runs now. If you notice, guildwars at most takes up 5kb/s while playing, even when mapping from different areas. When you sign on, it has to read the contents of your character's storage, your storage bank, characters names, armor sets, hair styles, NPC direction and more. Surely if they are able to do something like that with a minimum of bandwidth, an AH is possible. (Oh, and the models and textures that are being loaded arn't being loaded from the servers, they're being loaded your computer using what are probably identifiers that link to that armor set)
I think an auction house would be a great idea, if they could do it without being owned by blizzard.

Of course it is possible to make an auction house, WoW has one. An auction house is simply a database, and there are databases that are capable of serving a large number of users (and if there are bandwidth and computational issues, you can always solve them by limiting the number of items players can post, limiting search results to the 10 cheapest items that match, etc).

The biggest problem is making the auction house work with the game. Depending on how the game was programmed, that might or might not be feasible. Making an auction house might require many existing modules to be changed, and it simply would take too much work to change those modules and retest everything.

Think about it this way. Many games have a z-axis. Guild Wars right now is essentially a 2D game that looks 3D, you can't really jump or climb, etc. So is it easy to make a game with a z-axis? Yes, many of them exist already. Can you just add the ability to jump (the z-axis) into Guild Wars? That would probably require most of the code to be rewritten from scratch.

The problem might not be a technological limitation (with bandwidth, storage, etc), but with feasibility instead. You can't just rewrite the entire game (well you could, it's called Guild Wars 2).

I hate to tell you, but some of the computer science analysis is WAY off. You really should go read some database architecture books and packet technology books. I know you mean well, but the analysis from a technical standard point is amateurish.

You made a lot of assumptions, but many of them flaw your analysis. I won't be detailing every one of them but the most glaring is:

Assumptions such as 3mil active users ... is WAY off. I will wager the max number of active users is closer to 1% to 2% of the 3mil users.

updates on bids is only essential when u actually look at your bids, so a "get" function will only be performed when bidder will be checking wares, not wehn someone else starts bidding at it. The assumtion by the OP that 0.081% of 3 miljon copies would result in that many items is ungrounded. Many people have multiple copies, so will use only one account for trade, many accounts are just hopelesly abandonded, and many have been permanently terminated.
if the amount active would reach 400.000 to 500.000 I would still be surprised.

As to that people would use actively an in game auction system to get that bit more golfd from scrap goldens then yeah then I probably agree with that. You could limit the amount of space that can be used for auction, maybe opnly one or two items at the time, scrap will be low on the priority list, and probably just merched.

I realize that it will take quite a database, but saying its near impossible is well nonsense as well. Not all people do simultaneously querry the base and a good architecture will stream the process. Together with a limitation on the amount of auctions it can be feasable.

I never said any of this was possible and I never said that this would be the best way to do it I said it was a simple way I came up with on the supr of the moment to show the magnitude of things that would need to be considdered when implementing such a thing. and just about every last person who has replied has done nothing but support my statement that there is much to think about and considder when doing such a thing no matter what system they implement there is always going to be someone who says lets do it this way or could we change this wich is EXACTLY what most of the replys here are looking at the system I set forth and stated how it could ber changed improved or altered ALL of wich would have to be done by A-net themselves when designing an AH for us in game perhaps you can now see the point of this thread wich was to display the many things involved with implementing the AH alone

Even though I am a member of the Guru Auction site, I would use an in-game auction more readily because of convienience and automation.

Wow.. that was.. Well researched and informative.
Telling him that he's doing an amature job then making a guess on %s doesn't help your argument.

This is a repost of what I said in a previous thread. Using binary storage instead of just "digits" dramatically reduces storage requirements. And don't fall into the trap of wasting bits by storing discreet data into separate "bytes". That's not how programming is done.

Here's my original post:

The game already has a database of all the items. How do you think it's able to keep track of our inventory and storage? How does it know which items each character has? There is already at least one database that has every item in the game.

What is needed is a schema update that adds several fields to the item database:

1. For sale or not (binary value)
2. Reserve price (24bit would allow up to more than 16million gold as a price)
3. Current bid (24bit again as above)
4. Auction duration (if auction is in hours, an 8bit value would allow 256 hours (or up to more than 10 days)

A couple of interfaces would need to be built:

1. a way of marking an inventory item as being for sale and setting fields 2 and 4. You could even mouse-over the item to view current auction status of that item.

2. an interface that filters and displays the entire player item database by the first field (whether an item is for sale), provides additional filtering and/or search options, and provides a way of entering bids (updating field 3)

Lastly, some code logic would be needed (probably the only part that would require any real effort) to close the auction and transfer the item to the inventory of the bidder. Adding bidder history or email would be more difficult (would require an entirely new database), but it not needed for a functional auction system.

It's not a question of technology -- most if it is already there.

Note that the additional fields would add 7 bytes and 1 bit to each item in the database. Let's say 8 bytes to make it easy. While adding 8 bytes to each item in the database scales up to hundreds of megabytes of additional storage space GLOBALLY, its a small fraction of the amount of information being stored about each item at present.

And this is probably not how items are stored. There are a finite number of item combinations, and it might be more efficient to manage them by assigning each possible combination a unique ID. Depending on the number of combinations, this ID might only be a few bytes (4 bytes gets you in the billions of combinations). If this is the case, the database storage requirements might even be lower.

-Forjo

rofl


Anyway, although I like the concept behind WoW's Auction House, I think that a private store system would work so much better in Guild Wars. Not only very convenient, but easy to implement. Of course I'm not assuming anything with full certainty, but darkorical's post proves it in a way. There's no need to create a fancy database, or huge traffic transfers. You would simply open up a small shop with 4/5 items in a designated area for that purpose.

The problem of overcrowding would be countered with the district system.

One other thing, someone here said that if an improvement implied a rewrite of a complete game, that would be GW2.
Although you are right, we shouldn't look at things this way, for the sake of the consumer. What I'm trying to say is that, as long as I see Guild Wars chapters for sale in video games stores, I consider the game alive and well. If it "dies" let A.Net remove every box out of every store. Ridiculous, right? So why should they dump the support for a game that is still being bought?

I don't think they have any excuse whatsoever to say that they have bigger priorities now. As the good company they are, they should be prepared to support their current titles, *and* work on their future releases.

Cheers.

Seeing how ANet messed up something as simple as PvP skin unlocks by making them char based, I now think that an auction house is completely infeasable, not only in GW1 but in GW2 and probably GW3 as well.

1% to 2% of 3 mil active users is much much closer to the truth of 3mil active users. It is an assumption, you are right.

Anyone worth his salt in packet and database design will see that there are numerous errors in his original article. The OP meant good with the post so I'm not here to blast him.

1% of 3mil users is obviously wrong. Thats only 30,000 players, and you'll see that many if you pop to Lions Arch.

Forjo's Post makes the most sense to me. Since there are already item identifiers present when we see our storage, it makes sense that the only things you'd have to add is an auction interface and a restructuring of the item object's structure, as well as code to transfer the item to the bidder and money to the biddee.

A full district only contains about 250-300 users... 30,000 sounds right to me.

I think u cannot see 30000 people pop up in one region as they fix the amount of users in one region to much less 100-200 maybe.2-5% isn't so bad for an estimate maybe, considering that there are a number of world regions, and in those region the three main cities harbor some 5 districts. I dont have the exact numbers, 30000 maybe a bit on the low side, but if u would guess 50000-100000 I would totally not be surprised

That math is wrong. There are not anywhere near 3 million users. Most of the 3 million copies sold are the same people who bought Chapter 1 and then linked Chapters 2 and 3 to that.

I highly doubt there would be any legal issue with an auction house. Many other MMOs already have them (and WoW was not the first) without any legal problems. The only possible issue would be if ArenaNet literally copied code from Blizzard, which has almost no chance of happening.
Most of the infrastructure is already in place with the NPC trader system. The underlying design would need to be fleshed out quite a bit, but we already know it's possible for the game to query a database when you interact with an NPC and return a list of items for sale. It's not nearly as large of an engine change as the addition of a z-axis would be.
To add to this point, the recent Xunlai storage update increased the data representing our storage by a factor of four, whereas adding 8 bytes probably wouldn't even double it.


Also, I was surprised to see resistance to the idea of having non-perfect items in the auction house. Sales of near-perfect items would be a great benefit to casual players, who could quickly and easily get +28 or 18/20 mods for very low prices. It would benefit casual sellers as well, who normally would have a hard time finding buyers for their "worthless" items... they could at least get something better than merchant price.

And it's not like the junk items would bother hardcore players... it would be trivially easy to implement a filter so players could choose whether they want to see only perfect items.

1% is actually a good estimate. Even if there are 300,000 users playing (vastly overestimating) at any given time, not all of them will be using the auction house. Maybe only 10% of the 300,000 is searching/posting on the auction house. Storing and transmitting the items on the database is very cheap. Having large number of people query the database at the same time is not.


I'm not so sure that the NPC trader system uses a database. The NPCs only keep track of a very small number of items, so they only have to keep track of quantity, price, and some sort of variable for demand (which controls how prices change). A true database has many other options, primarily sorting and filtering (i.e. Out of the 50,000 listed items, give me all Chaos Axes. Or give me all Chaos Axes with +5 energy. Give me all the items listed by user X. Sort by price. Etc.) I don't think the NPC trader system uses all the functions of a database.

Alternatively, ANet could make a market town with more specialized NPCs, such as an inscription trader for weapons, tomes, etc. It would at least make trading for some items easier, and it would make a more centralized place for trading.

Well Im gonna be away for a few weeks. you guys keep right on debating this tho and we'll just see who can perfect the auction house sytem first you or A-net remember you guys will have to find a system that fits exactly into EVERYONE'S needs will work PERFECTLY on EVERY POSSIBLE SYSTEM does not leave anyone out and does not change the existing econemy, game play or other wise impact the game itself.

good luck

on another note If A-net is too slow to make an auction house why don't we make a live auction. Wednesdays and Saturdays at Great Temple of Baltazar ID1 X person will be designated as auctioneer use local chat and hold live bidding. Item entry from X time till X time all items given to auctioneer and money paid to auctioneer. Auctioneers could post all IGNs here along with contact info. they would screenshot all transactions that way if anyone complained of scamming they could defend myself with the shots and if they really did scam A-net wouldn't have trouble tracking them down to ban them

A recent interview was done with ANet Execs, and they talked about the feasability of an auction house. See it here:

http://www.dbms2.com/2007/06/09/the-...of-guild-wars/

I've read through most of this post, and though I admit I don't understand everything written, I have a possible solution that might allow for an auction house. The problem (even here on GWguru) seems to be the sheer number of items up for sale.

It seems this could be significantly reduced in an auction house setting if ANet "charged" 10% or 1 platinum (whichever is more) to post an item in the auction house for sale. This would keep people from posting all chalk, grapes, and most golds. An item would only be worth posting for sale if it were worth more than 1 platinum.

Also, there should be a time limit on items, maybe one week. If it didn't sell in a week, the original owner got it back. If they wanted to re-list it, it would be another 1k or 10% of list price.

The auction house could further be simplified by not allowing competing bids. In such a case, it is no longer an "auction" house, but simply a "store" but this would still allow a place for people to sell stuff.

One drawback to this is that it would probably not cut down on the amount of spam in game. People would still be spamming "W T S 34 wood planks 500g."