Wobsot is an experimental forum designed with the goal of remaining interesting no matter how large, diverse, and stupid its user base becomes.[1] It has a number of unique features and attributes to accomplish this.
An informal IQ/curiosity test at the front door. When a user first comes to Wobsot, he is greeted with a page that has been carefully designed to be as unwelcoming as possible (including the phrase “Welcome” at its top). Users are given no hint as to what the site contains—all the front page has is a username/password login and an “I’m Feeling Lucky” button.
New users will be unable to log in, so they may click the “I’m Feeling Lucky” button, but the page refreshes, unchanged. The “I’m Feeling Lucky” button is actually the mechanism for creating new accounts, but no hint of that is given. Furthermore, undocumented requirements on username and password length (3 and 8 characters, respectively) reduce the likelihood that people will get in through dumb luck. The entire front page is a challenge with no perceivable feedback or reward.
Information about the site itself is tightly controlled, and there are even some red herrings. Besides the active controls for the site, there are only 3 static pieces of information given on every page: The word “wobsot,” which is made up and meaningless; the subtitle, “Is this thing even on?”, which conveys general confusion; and the footer, “2005,” which looks like a date but has no significance at all. (It certainly isn’t when the site was founded.)
No indication of site size.[2] No count of messages or users is ever displayed. Random lists of messages update infrequently (currently every hour), so you have to be patient if you want to judge the site’s size by how frequently specific entries appear. Usernames aren’t displayed, so it’s impossible to know how many other authors there are. Even in the source, messages are identified by hashes instead of indices.
Very limited navigation. There is no site index. POST is used exclusively instead of GET, so users cannot create arbitrary URLs. All of the message lists (besides the one on the user’s personal page) are short and infrequently randomized. There is a search field, but it is a secondary function of the message creation field—and it is difficult to tell if a search query (AKA new message) existed before it was entered. Because the site uses POST, the back and forward buttons in the browser, not to mention bookmarks, don’t work. Users can “claim” messages, which is equivalent to having authored them and includes them on the user’s personal page (a substitute for bookmarks).
Navigating segregates by interest. Each message has a small list (currently 5) of softlinks, which are connections between messages left behind by other users browsing the site. Some of the softlinks (currently at least 2) are chosen randomly from paths which have never been taken, and the overall order is always random, to encourage churn and discourage intentional softlink creation. Because softlinks are ranked per-message, only users interested enough in a message to find it are able to “vote” on its softlinks. As users follow softlinks, they are lead to pages that were found interesting by others who came the same way. There is no reward for, or even indication of, softlink creation.
All messages are uniqued. This prevents repetitive content from flooding the site. (And since POST is used for everything, there is a high chance of users inadvertently double-posting.) Unlike ROBOT9000, uniqued content is still kept, which should further discourage its reuse over the long term. The site uses a monospaced font, which, besides being vaguely unfriendly, also prevents common letter substitutions, such as an uppercase “i” for a lowercase “L”.
Messages are reference counted. By writing, searching for, or explicitly “claiming” a message, users increase the message’s reference count. Because claiming a message puts it on the user’s personal page, which is the easiest way to find it again, claims will be a reliable explicit indicator of a message’s popularity. This is a side effect of message uniquing and bookmarking—reference counts are currently unused for anything besides message deletion.
The users are de-emphasized. Users besides the one currently logged in are never displayed.[3] As well as hiding the number of users, this prevents a lot of drama. Even the logged in user’s personal page has no room for personal information—it merely serves as a list of claimed messages.
Lurking is discouraged. Because an account is required to even view the site, and because claiming messages is the only way to bookmark them, it is difficult for users to passively observe the site without changing it. However, although contributing to the site is encouraged, there is little damage that new users and idiots can cause.
Links to other sites must be manually copied and pasted. This prevents sites which are linked to from using the referrers, and discourages users (and spammers) from posting and following worthless links.
In conclusion, Wobsot’s subtitle—“Is this thing even on?”—very effectively represents the user-hostility and incomprehensibility it was designed to embody. In the unlikely event that it builds a user base at all, it may just succeed in its mission to prevent its users from destroying it. Only time will answer the question, “At what cost?”
Addendum: In the time since this article was first written, I have decided that Wobsot is built on a failed hypothesis—albeit not the one described above. Its flaw, like Twitter’s, is that it does not provide much support for reasoned discussion. Its messages must be short and are effectively one-way. This makes it boring.