History Flow Shows How Wiki Articles Evolve

teslatug writes "IBM has released a preliminary alpha version of its History Flow Visualization Application that shows how collaboratively created documents evolve. The tool is written in Java and it's available for download along with plugins for MoinMoin and MediaWiki. They have some interesting screenshots of the Wikipedia articles on abortion, Brazil, and love."

Comments

Here's an Idea

[great throwdini]

Instead of linking simply to the download page and the screenshots, give people a chance to RTFA and link to the History Flow Visualization Application's overview document.

Huh?

[Anonymous Coward]

The screen shots are so poor they look like some geological data.

Can any tell if there is going to be an earthquake soon?

Interesting but useful?

[thundercatslair]

This program is interesting to look at for a little while, but can it do anything useful? I don't really see a need to see the history of a wiki visually.

Re:Interesting but useful?

[shadowmatter]

From the Overview page on alphaWorks:

The patterns revealed by History Flow Visualization show such information as spacing by date; occurrances of vandalism; authorship; growth; and persistence.

It seems like a good tool for inspecting the history of a document at-a-glance, but you're right -- for more details, there is no substitute for a commit log.

Could be useful, however, in environments such as CVS or Subversion across sets of files... Hmmm.

- shadowmatter

Heavy Metal Umlat

[hatrisc]

Heavy Metal Umlat is a very interesting look at the history of a Wiki page. Worth checking out.

Re:Heavy Metal Umlat

[Raul654]

I had the great pleasure of showing that video to David Gerard, the principal author of Heavy Metal Umlaut. He was floored, and thought it was the greatest thing he ever saw :)

As much

[odano]

As much as I love visualizing things that aren't visual, it just doesn't seem like this application changed the data into anything useful.

I have no idea what the evolution of those documents was before, and even after viewing the visualizations (and knowing what they mean), I still have no idea what it means about the document.

Re:As much

[FarHat]

Without the actual document, a graph such as this doesn't tell you anything. What it could tell you, along with the document that it is representing would be much of a document changes in any given time. Are there parts of the document that are essentially static. Parts that are static would be parts that there is little disagreement about. Parts that change a lot could be considered controversial. Heavy editing would indicate a lot of popular interest in the article, etc.

Re:MOD IBM -1 REDUNDANT ;-)

[Stalyn]

poor Amsterdam Vallon... his posts are somewhat informative and interesting but are constantly being modded down because he asks people to think of Terri Schiavo when right now a lot of people can't stand her (well the media coverage).

btw Amsterdam if thats your real name are you of any relation to Archimedes Plutonium?

svn blame

[TrdrJoe]

Tools like "svn blame" or "cvs annotate" are much more useful; they tell you who added each line of text in your file, when they checked it in, etc.).

Still, these tools don't let you see the history of text that has been *deleted*. A visualization like "historyflow" could be useful there

Wondering what "Moin-Moin" means?

[Doomie]

It's how people greet in Northern Germany (especially Bremen). I guess it comes from "Morgen" (morning), but people say "moin-moin" pretty much all the time.

Thank you for your attention :)

Information visualization is tough

[shanen]

You're addressing a difficult problem there. You have to decide which dimension you want to consider, and then you have to provide data that spans the dimension in a meaningful way. The samples given were apparently picked mostly for their heavy activity resulting in "pretty pictures", but that isn't a particularly relevant or significant dimension.

My own interest would be in visualizations that identify zealots of various stripes violating the basic neutral POV philosophy. Something that would show the behavioral similarities in their behavior. I must be too interested in deviant behavior? For example, there was some recent ruckus about the "online poker" entry, where some commercial zealot was trying to use Wikipedia as free advertising to flog his poker Web sites. Before that, I remember a similar incident involving a religious crazy who wanted to use Wikipedia to manufacture some credibility for his cult. I'm sure there must be some tranplanted Newsgroup Charlies wandering around Wikipedia, too. (Don't look at me--I'm just a harmless grammar Nazi.)

In practical terms, if you can identify patterns associated with such problematic behavior, it will make it much easier to create automated alarms to help people notice. However, I'm kind of skeptical about the idealistic approach of trusting people's common sense. I'm given to understand that the Simpsons is a popular program, but it is so profoundly anti-intellectual that I can't stand it at all. Then consider some of Dubya's knuckle-dragging supporters and their primitive belief systems...

Never underestimate the power of organized knuckle-dragging.

Visualisations as writing tools

[jesterzog]

These visualisations are quite neat. I've often wanted a word processor that would be able to do something like this. I tried writing on a private wiki at one point, but it still presents the changes between different versions very separate and discrete from each other, and from the editing, so it didn't work terribly well.

When I write things, the text often evolves a lot over several days. I usually blurt out everything I want to say at the beginning, and then go back and edit it over and over again until it's expressed how I want it. One problem, though, is that when I go away and come back again, it's not always obvious which parts are the most volatile, and might need the most attention. It often takes a while to get back into the right mode of figuring out where the complicated parts are, and editing the document.

Writing on paper is still very different from a word processor. It's very obvious where a lot has been crossed out and changed over and over again, and previous crossed-out versions, even if they're on paper that's been put aside, are often still visible and easily accessible during the rest of the process. In a word processor, though, nearly all of this contextual information is lost. At best it's possible to "track changes", and that particular tool is relatively simple and usually aimed at being able to see some one-off changes that someone else has made to your document.

Beyond just tracking changes, which is a very linear representation, I'd love to be able to have some kind of visual representation surrounding the text to indicate the stability of different sections of what I've been writing.

Some useful ideas might perhaps include different coloured backgrounds to represent the volatility of sections of text, blocks of text that get moved a lot, being able to quickly flip back to what a small section used to be (without necessarily committing to it), and so on. Perhaps even a draft mode that shoves text aside (maybe above or below), but still leaves it accessible while editing the replacement text.

As a writing tool, it'd be a very helpful extension to any of the open source word processors out there. I bet there's a great niche market in authoring tools that current word processors really don't cater to right now.

Troll metrics

[Tablizer]

One step closer to objectively identifying who the wiki trolls are :-) What color of lines do they use for trolls? Toad green?

It's based on diff format...

[vhogemann]

So it might be used to show progress over time on open-source projects. It would be usefull to show progress over a single project or how two projects merged, and to show wich contribuitions made it to final versions, or witch developer has more code on it.

It should be very interesting to see it applied to big projects, like the Linux kernel or the KDE project to see how it evolved from the number of contribuitions and devellopers, and to see how long each contribuition survived unnaltered on the source.

It could prove to be a very usefull tool indeed.

The author

[rexguo]

Martin Wattenberg, being the main author, also has a personal homepage that has very interesting visualisations in Java as well: Bewitched

Ironic that this gets posted on Slashdot...

[corporatemutantninja]

...yet Slashdot passed on an opportunity to have something like this for themselves.

The IBM researcher who created this software, Martin Wattenberg, also wrote some really cool tools for visualizing and navigating Slashdot threads. He said he would be happy to let Slashdot use them for free so I made an intro but the /. guys never followed up.