tsunami
log in
email
password
links
newest items
tag list
syntax reference
tag:time
history
pointed to by
TIME Forums
Ian notes on tsunami
add
item name
tags
==Goals - maintain the ease of use most forums have (actually, increase it) - facilitate the creation _useful_ knowledge ==Knowledge I want to maximize the utility of information contained in forums. Many forums do a sufficient job of allowing users to create threads and posts. I am unaware of any which help one condense and solidify information discussed. Consider two situations where the simple, thread-post paradigm breaks down: =Problem & Solution Consider a typical thread in which the author presents a particularly perplexing problem. Tens or even hundreds of posts later, the problem is finally solved in its entirety. If someone comes along later with a similar problem, he/she will probably have to read most of that thread, skim for text where the author elaborates on the problem, ignore off-topic posts, and piece together a solution that most likely spans multiple, non-consecutive posts. It is my opinion that as the number of readers increases, _time spent condensing the solution << time saved reading the condensed solution_ (over the solution that is effectively spread over many posts). I want the forums I write to facilitate this. =Debate Consider a particularly intense debate, where tempers run hot, insults are made, and tangents abound. Determining which post was in response to which can be tricky with flat forums (as opposed to threaded [multiple levels deep]). Filtering out the irrelevant material can be quite distracting. Would it not be useful to see exactly which text, in which post, a given response is directed towards? Would it not be useful to look at a given post and see which parts of it evoked responses? Currently, the only way to solve this is to scan through a potentially large volume of material. This is undesirable. ==My Vision Some forums make it easy to quote previous posts -- along with the _reply_ button at the bottom of the page, there is a _quote_ button for every post. Unfortunately, quotation functionality is not always used and when it is, it can be distracting and/or inaccurate. For example, quoting the post directly above simply adds redundant text that slows the reader down. Then there are those who quote another user but "correct" what he/she said, making it impossible to perfectly match the quoted text with the source. What if the actual post ID & character indexes of the quotation were saved? This way, not only can I find exactly what was quoted from the later post, but I can look at the original post and see which parts were quoted, and by whom! Add versioning of posts, convince users to edit their posts in response to others, and one can all of a sudden get strong linking between what one person says and what another says. The cost is minimal: the plaintext version of quoted text would include four extra pieces of information: - ID of post that was quoted - version of that post - beginning character index - ending character index This information would all be automatically determined by software. Ideally, one could gather text-to-be-quoted while reading a thread by selecting bits and pieces, then have access to a "quote bucket" during a reply. Next, consider the ability to tag a thread, post, or even section of a post. This would enable, among other things, the ability to find all discussion in a thread that is centered around some sub-topic/tangent (if, of course, tags were used). =Notifications Three core features have been suggested: - quotations that link strongly to a post, version, and character indexes - versioning of posts - tagging of threads/posts/parts of posts Using this information, we can enable advanced notifications not found in [most] other forum software. - when a post is edited - when a post or part of a post is quoted - when something is tagged In a time where information overload is becoming a colloquialism, the ability to filter what one sees is essential. The criteria listed above allow for more powerful filtering than I have ever come across in forum software. =Search All of the information above can be used to allow for very powerful search. In my experience, forum search software is woefully inadequate. It is usually worse than Google search, hearkening back to the Altavista/Yahoo era, an era to which I do not wish to return. Through the use of powerful text searching algorithms (perhaps via "Lucene"[1]) and tags, I hope to provide a much better search experience than users are used to. [1]http://en.wikipedia.org/wiki/Lucene ==Organization The idea of maintaining separate forums and subforums has never sat well with me. Sometimes they are well delineated (like feature requests vs. sales questions), but oftentimes they are not. Operators of busy forums tend to split their forums into enough pieces so that no one piece: - gets busy enough to make new posts disappear too quickly - is underused and ultimately abandoned I consider this an [[Thoughts on Information Classification/126:information classification problem]], to which my answer is tags & [[TIME]]. Like the granular quotation system described above, I want to be able to apply tags to entire threads, single posts, and arbitrary characters in a post. Some threads won't use this feature at all, but I am hoping others (especially large ones) will use it extensively. ==TIME Integration Many words have multiple meanings. If the domain being discussed is small enough, this is often not a problem. However, with the amount of information on the Internet, this often _is_ a problem. Fortunately, "Google"[1] has done an excellent job trying to match the different terms input to return the most likely result. Google uses the "PageRank algorithm"[2], which results in the most "relevant" version of any set of terms will be returned. Unfortunately, when the thing you were looking for with terms _X_ and _Y_ is preempted by something considered more relevant with the same terms, you are out of luck unless you can find a term _Z_ which filters out the "more relevant" results. [1]http://google.com/ [2]http://en.wikipedia.org/wiki/PageRank Another problem is that even when you have a good list of results that probably match what you are looking for, there is no way to categorize them, to find out _how_ they discuss the term you are interested in (without actually scanning their contents). [[TIME]] was created to take knowledge management to the next level. When tags become insufficient (due to the problems noted above), my solution is to create TIME items and replace tags with links to those items. This solves both problems above: disambiguation of like/identical terms, and [optional] documentation of why a thread/post/part of a post is linked to a TIME item. (Remember, the link between two TIME items is actually a third item, which can itself have tags and text.) ==Questions - What sort of markup should be allowed? A "lightweight markup language"[1]? [1]http://en.wikipedia.org/wiki/Lightweight_markup_language ==Links - "Jeff Putz's blog on writing POP Forums"[1] - Jakob Nielsen: "Write Articles, not Blog Postings"[2] [1]http://weblogs.asp.net/jeff/archive/tags/POP+Forums/default.aspx [2]http://www.useit.com/alertbox/articles-not-blogs.html
some permissive license goes here
contact