Yax over at Dungeon Mastering recently put together his list of the top 50 (now 52) RPG websites out there. Coming out of nowhere, Obsidian Portal shows up at #13. Of course, he readily admits that his rankings are totally biased, but that makes it worth even more to me. Rather than trying to quantify why Obsidian Portal is any good, users just like it.
Anyways, thanks Yax for the link and knowing that we exist!
There are a handful of websites that have sprung up to help gamers find each other. It’s a great idea, and was one of our original motivations for creating Obsidian Portal. At the time, I didn’t know about the other player matching sites, and I was too scared to ask, thinking someone would steal my awesome idea. 
Unfortunately, most of these sites just don’t have the critical mass necessary to make them useful. I have signed up on a few, and I have never received any sort of response. Seeing as how I live in Atlanta, that’s a bad sign. If I can’t get matched up with players in a big city, there’s no way people in rural areas are getting any use out of it.
What if there were a way to combine and aggregate all the data from all the sites into a single index of gamers? In this day and age of web services and mashups, there is no reason that this should be impossible. So, for your consideration, I am proposing a communications protocol that will allow all the player matching sites to interoperate for the purpose of providing more results to their users. I am calling this the Lonely Gamer Protocol, or LGP.
Update 2007-11-06 I’m looking at using Google’s OpenSocial as a possible alternative to Atom. OpenSocial’s people element has almost all the necessary fields, and the gd:ExtendedProperty might be sufficient for the rest.
Of course, the main motivation is to provide a better experience to our users. By sharing our data with each other, we will beef up the number of results returned to our users, thereby greatly increasing their chances of getting a hit.
However, at least in my case, there is a secondary, more commercial motivation. With Gleemax breaking onto the scene, Wizards of the Coast has become the 800 pound gorilla in the room that none of us small timers can compete with in terms of marketing or exposure. Unfortunately for the users, Gleemax just isn’t that great, at least not right now. They have a lot of promises, but very little in the way of actual features.
If we all try to compete directly with Gleemax, we will probably fail. They have an established userbase and a war chest they can dip into for advertising and paying their employees. However, if we band together and attempt to make interoperable services, we’ve got a shot. Pretty much any of the current gamer-finder sites is already 1000x better than Gleemax, and working together will only strengthen that lead.
Of course, Gleemax is free to support LGP as well, and this would be a great boon to the entire community. However, given its current state and their speed of resolving issues and feature requests, I don’t see this happening for a looong time.
Please comment below if there are goals I should add or remove. Please note, however, that projects that are able to condense their goals down to a short list do better than ones that throw in everything and the kitchen sink.
LGP consists of a communication protocol along with a simple REST API for retrieving the data from participating providers.
The actual data involved in such an exchange should be very simple. Shooting from the hip, here are the fields I could come up with:
XML is the obvious choice for the data exchange. That’s a no-brainer. Further, in order to further lower the barrier of entry, I suggest that we use a preexisting XML protocol as a base layer. Most of the fields I’ve suggested could be easily stuffed into an RSS or Atom feed.
| LGP Field | Atom Field | Comments |
|---|---|---|
| Name | title | This would be the username, nickname, or campaign name |
| Geographic Location | georss:point | We can use the GeoRSS standard for Atom feeds to encode the user’s location. It would be encoded as a simple point element. |
| Preferred Games | category | Cram all the games you want to play in as category elements. For bonus points, we can define a closed vocabulary for the more popular games. Of course, this would be in addition to an open, free-tagging scheme. |
| Looking for game | ?? | A flag to indicate that a player is looking for a game to join. |
| Looking for players | ?? | A flag to indicate that an existing group or campaign is looking for more players. |
| Link to profile | link | This would be a link to a profile page for the user. Following this link would be the primary way of finding contact info for the player. Note: This link would probably also be used for the id field required by an Atom feed. |
| Last update / login | updated | This field indicates the last time the user updated their profile or logged in or something similar. The main goal is to let the stale entries settle to the bottom while the active, searching players can be easily found. |
Looking back on that list, there are a few fields that need to be fleshed out and some formats need to be agreed upon, but the core functionality easily fits. Using Atom as the base for this protocol provides a lot of great benefits. As I mentioned before, most languages will have an Atom parser, making feed consumption that much easier. In addition, many will have tools to help construct feeds. Finally, it will be possible for the data to be made available to individual users, if they wish, just by subscribing with an Atom feed syndication tool. It might strip out some of the additional fields that we add, but if you provide filtering at the upstream level (ie. “Subscribe to a feed of all the gamers within 50 miles”) then it shouldn’t matter.
I’ll be the first to admit that I’m not really sure what’s required here. There needs to be a way to harvest the data from the individual sources. Further, there will need to be a way to filter the results, both in order to save on bandwidth, but also to make the results useful.
The following parameters allow for filtering based on set size, preffered games, and the last time a player’s profile was updated.
| Parameter | Required | Comments |
|---|---|---|
| limit | N | The maximum number of entries to return. Defaults to something reasonable, like 100. |
| start | N | The index of the first record to receive. Manipulating start and limit allows a client to iterate through a result set. Defaults to 0. |
| categories | N | A comma separated list of categories. This allows for filtering based on which games a client is interested in. |
| last_update | N | A date field (what format does Atom use?) indicating that you only want results that have been updated after the specified date. |
Filtering on geographic location would probably also be incredibly useful, but it’s not nearly as trivial as the previous parameters listed. Some tools make geocoding and “within X miles of Y” type searching very easy. Still, this would raise the barrier of entry. Perhaps it could be an optional filtering scheme that individual sites could implement, but was not required in order to be compliant with the protocol.
Probably the largest hurdle to LGP would be getting the individual sites to share their data. For the larger sites, their userbase is their most valuable asset. Making it freely available might not seem in their best interests. However, it’s obvious that from the users’ perspective, LGP makes a lot of sense. Once the user data becomes a commodity, the sites will be forced to differentiate themselves based on their other features. For Obsidian Portal, that should be easy.
Still, joining up and sharing your data should generate a lot of traffic back to your site, and therefore signups or purchases or whatever else you count as a conversion. This is because the primary way of contacting a user will be through their profile page, which most likely exists on the site that provided the data. You could even require that someone sign up for your site before they can contact anyone, although that might be sort of a jerky thing to do.
Everyone implementing LGP as a provider would need to make this very clear to their userbase and give them the ability to opt-out. In fact, perhaps opt-in would be a better policy, forcing users to specifically elect to be include. Still, why would they sign up on a player matching site in the first place unless they wanted to be found? I imagine most users would jump at the chance to be included.
The main issue with privacy that I can see is related to the geographic data that is passed with the feed. Giving out your address on the Internet is generally considered a Bad Thing, and this is no different. To that end, here are a few approaches.
First, you could restrict the lat/lng to a city area only. Therefore, you pass nothing more specific than a user’s city. The obvious downside is that for a city like New York there would be thousands of identically located entries. If you tried to map them, it would end up a mess, with all stacked one on top of the other.
Second, it might be possible to include an intentional error into the geographical data. Instead of feeding out the user’s exact coordinates, the system could give a random point within X miles of the user’s actual location. This would effectively mask their exact location, but would probably be very complicated to implement. Further, users would jump around on a map every time it was refreshed. Not a great solution.
Finally, the solution I propose is to pass the onus of anonymity on to the users. When asking them for address info, make it clear that they are not to enter their exact, home address. Instead, ask for nearby landmarks or street intersections. The Google Maps geocoding API is good enough to translate such strings into lat/lng coordinates. This would allow the users to accurately place their location, while at the same time assuring their anonymity.
I am already signed up on multiple sites, and I imagine there are others as well. So, if all the data were aggregated, I would appear multiple times.
For now, we’ll punt on this issue from the LGP perspective. Individual consumers are free to try whatever de-duping mechanism they think will work best, or just forget about it. I imagine that even very dumb de-duping (ie. throw away all hits with the same name within 1 mile of each other) would be pretty good. Plus, unless people routinely sign up on 50 different sites, it won’t be a huge problem.
Ultimately, it would be nice to have a central registry that assigns unique identifiers, but trying to establish such a registry would probably be more work than it’s worth at this point.
I’d love to hear feedback from anyone interested. Protocols like this only succeed if they achieve wide adoption and serve everyone’s needs. So, if I’m unfairly excluding someone, or I’ve missed something from the main list of goals, let me know in the comments. I will update this post over time as we come up with new ideas.
By no means is this document meant to be an academic discussion or pointless pontification. This is a very rough draft of what I hope will be an actual service that users can interact with very soon. So, I will go ahead and make a challenge to my fellow site owners: Let’s finalize a spec for LGP by December 2007, and have it implemented on our sites by Christmas (Dec 25th). If we could help our users find someone to game with over the break, it would be quite a present.
For my end, I solemnly swear to have Obsidian Portal implement at least the provider side within 2 weeks of the spec being finalized. I feel comfortable making this promise because I know just how easy it will be to do. Heck, it will probably take less than 3-4 hours to do the actual programming.
To any actual gamers reading this, you have a job too. Go back to the sites that you’ve registered on, post some links to this article, and demand that they support LGP. Tell them that by supporting and participating in the network, they will greatly enhance the user experience on their site. Believe me, once the users start demanding things, it gets very hard to ignore.
Recent Comments