Thiblo - The Big Picture

Here we'd like to describe the "big ideas" behind ThiBlo. Not the technical stuff, but the focus points, the pure mind constructions which we think can make the big difference. The items are in no particular order.

These are pretty ambitious plans. You can also read a more concrete description about our first steps, and the way we approach the implementation.


WebMore is a novel spatially organized commenting system. It could have several important implications.

The fact that one have to select the sentence to comment may have a psychological effect. The necessary moment of thinking what to answer may raise the quality of comments.
WebMore bubbles are naturally threaded. The tree of comments is organized by topic and not just by conversational threads.
Spatial organization of comments is crucial for articles of larger volume. You have to be able to comment on the exact semantical unit you'd like to. Otherwise valuable feedback will be lost.

So, the goal of WebMore is to provide a commenting system which somewhat ease off the cognitive pressure on the reader of the comments and to raise the quality of comments. Combined with some /. style comment scoring system it may very well turn out to be an ideal commenting system. Or may be not. Time will tell. But we have to try to come up with something more useful then the current solutions.

Personal Newspapers

"The advent of the blogging era will terminate the traditional journalism. Citizen journalism will completely replace it." This of couse is not true, as has been pointed out by many intelligent people. (See ex.  this blog entry (in russian). Raw translation here.)

But. Even if the new era will not change dramatically how news are created, it will definitely change the way we are receiving these news. Already for many people the main source of news is their "Friends Page" at LiveJournal or some community-driven portal like Digg. So the world is changing to the state when everybody's main source of new information will be their social network.

Or again, may be not. But we believe firmly that this is a natural (and unstoppable) process. Some of us beleive that this is "the good way", and probably the only way to put an end to the current oligopoly of the news industry.

So, based on the LiveJournal's Friends Page feature, the current RSS proliferation, etc., we decided to make every Thiblo user to be an editor of his own newspaper. None of the aforementioned technologies are sufficient (for ex. quite a few LJ users wanted to be able to use several different "friends pages".)

Thus the goals (probably the main goals of Thiblo):

  • Make the information-flow in the social network to be as flexible and as useful as possible.
  • If an individual or a community want to create something more structured and influential then a blog, something more "newspaper-like", make it possible. And make it possible with little effort.

Social Metadata or the Role of the Community

The social information has many apperances in the current systems. Once again, LiveJournal's "friends page" -> "whom do I read"; "friends only entries" -> "who can read this entry" (not so important in my opinion). Or take an other noteworthy example, the community based moderation, Slashdot for example. On Slashdot we have one big community and everybody's vote is accounted globally. But for this to work one has to bear the trust of the whole community to be accounted. Hence the Slashdot "karma" and the race for it. Why do this information (who trust whom on what) have to be global? If I am a member of some little community (which I trust), I can take in account their vote with much more weight than the global moderation. (Note: Slashdot also uses this local trust/distrust information. Sort of.)

Or do we even need global moderation? Or just some number of big(/medium/small) communities which moderate for themselves. And the rules of different communities can be different. I can trust my closer friends unconditionally (and may or may not trust people trusted by them), but a larger group can use some sort of automatic karma...


  • make the social information as ubiquitous, flexible, and useful as possible.

Meanwhile, beware the complexity! Too complex system may turn out to be... just too complex.

Focal points for the first implementation

We'd like to create a journal-factory: real community journals (not only blog-like ones, but traditional newspaper-like journals with columns and edited or moderated frontpages).

content flowing freely
All content is created in the spirit of creative-commons. Articles are "put into the system", appear in some blogs. Blogs are like in/out channels: they can be given other blogs to collect from, and are subject to be used as a source. No intellectual-property nonsense.
semi-automatic flows, a.k.a moderation
An input-channel of articles can be moderated by the owner of a blog.
community has a real role
Being a friend means "content trusted by her is trusted by me", moderation is skipped
One needs to select what sentence to comment. We think this may going to have a psychological effect, the neccessary moment of thinking what to answer may raise the quality of comments. Webmore bubbles are threaded by nature. The tree of comments is not only organized by comment-reply-threads, but by topic.

Tags and content-flow

A blog, in other words a frontpage, is built from boxes of articles. A box shows the last few article tagged with a certain tag.

The owner of a tag can create rules describing conditions for the tag to automatically be attached to a certain article. This way we can have boxes being filled automatically from contents of other boxes. The neccessity of retagging each article introduces the possibility of moderation: a tag can be semi automatic, waiting for human OK after being attached automatically by a rule.

The users of have the possibility of defining boxes, tags, rules, styles. They can have real frontpages. The default is naturally a fallback to a classic blog: a new user has a default blog with one default box with one default tag.

Tagging in detail

We are using an "avalanche-like" model for tagging. We have rules, called aggregations. A rule is a triple (b1:A, b2:B, C), where b1 and b2 are blogs, A and B are tags, C is a boolean flag. If tag A is applied for a certain article in b1, then B is also immediately applied as if it was tagged with B in b2. This means that editor of b2 trusts b1:A as a valid source for b2:B and everyone who is subscribed to b2:B can transparently read b1:A (actually -- whatever meant by b1:A, it can be an aggregation too). C is the moderation flag (stop-flag). If it is true, then the new B tag is flagged as "to-be-moderated", and aggregations (B, X, Y) are not applied.

When the "to-be-moderated" flag is cleared for a certain tag, then new tagging avalanches are started immediately with all those (B, X, Y) rules.

Tags are not global, they all belong to a blog. This way tags are trustworthy as much as I trust the owners (decision-makers) of a blog.

Automatic tag creation is a good idea, however I consider it as an UI syntactic sugar.