As we plan for our second fedwiki happening the differences between federated wiki and wiki become, well, stranger.
If you’ve been following the story thus far, you know that federated wiki is pretty radical already. As with wiki, people converse through making, linking, and editing documents. But because each person has a seperate wiki, there is a fluidity to this “talking in documents” that is hard to describe.
You write a post referencing Derrida’s concept of “ultimate hospitality”. I get interested in that, do a bit of research. I save a copy of (fork) your post to my site, but link it to a page describing Derrida’s hospitality in more detail. Not because I’m an expert, but because it’s a good excercise to understand your writing. You see that post and fork mine back. The next visitor to your page finds your page plus my article annotating it. Maybe they edit it, creating their own fork. And so it goes.
Carry It Forward or Clean Slate?
As we move into Happening #2 on Teaching Machines, this question comes up: do we pull these documents into the new happening? Use the same site? Do I create one big “Megasite of Mike” which I haul into each event like my persistent twitter feed?
And what we’re thinking is maybe I don’t. Maybe Happening #1 site is done, and I create another site.
So for instance, my wiki farm is at hapgood.net. For Happening #1 on “journaling” I had journal.hapgood.net. Maybe for the Teaching Machines happening I make machines.hapgood.net. And to the extent I want to talk about something from a previous event in this new context, I fork it into the new context.
It’s pretty simple to do this in Federated Wiki after all. I just drag the page from one site and drop it on another, I edit it for the new audience, or maybe even take the opportunity to clean up a few typos. And that’s it, done!
While this may not sound extraordinary, it is in fact an inversion of how we usually think of wiki (and sites in general). And it has some neat ramifications.
The Dreaded Curve of Collaborative Sites
To understand why this is such a departure from traditional collaborative sites, we need to introduce you to the dreaded logistic curve. Here’s the curve of Wikipedia production:
Via Wikimedia Commons.
Until about March 2007 many people thought that Wikipedia’s growth was exponential. Above, we see a log scale graph, where exponential growth would be represented by a straight line (explanation of log scale).
But things go a little wonky in 2007. In that year it begins to become apparent that a logistic model might better predict Wikipedia growth. Currently the site is growing faster than a logistic model would predict, but well under earlier exponential models.
“Enwikipediagrowthcomparison” by HenkvD
If the extended model above fits, Wikipedia will near heat death in about 10 years.
Projection by HenkvD
People have attributed this to a lot of things, and certainly it’s a phenomenon with many inputs. But the logistic-like nature of it suggests one simple explanation — limited resources. Logistic curves are what you find when you map animal populations, for example, coming up against the resource limits of the environment.
In wiki, as in many other collaborative projects, the limited resource is stuff left to be written. People get together to write articles, and initially each new article leads to two other articles, and you see that growth on the left side of the graph.
Eventually, though, the easier stuff is written. It may not be written the way you’d like it to be, but it’s written. Wikizens move from homesteading new territory to vicious fighting over already cultivated plots of land. The enterprise begins to feel less fresh, and the type of people who find this stage envigorating are often dreadful bureaucrats.
Now, you might think this was just a function of Wikipedia, as it has articles on everything. But you’d be wrong. Smaller wiki communities see this same pattern. Here’s a graph of the first wiki approaching its own heat death:
(Chart by Donald Noyes)
Now these are the tail-end days of that wiki — it started in 1995. And its this observed pattern that led Ward Cunningham to predict, much earlier than most, that Wikipedia would follow the same pattern.
We get why Wikipedia would hit a resource limit — there’s an article on fish fingers already, for crying out loud. But why would WikiWikiWeb run out of subjects? There’s only 30,000 pages, after all.
The answer’s simple — people come together for a reason, a shared interest. In the case of WikiWikiWeb it was to share software design patterns and agile programming techniques. As those subjects fill out, the opportunities for new contribution diminish. The community has a choice — expand its charter (which fractures the cohesion of the group) or move into maintenance mode. (Likewise, Wikipedia could expand its charter by loosening notability requirements, for example, but this would radically change the nature of the project people thought they were contributing to).
Very few people want to be on a site that’s all about maintenance. As we mention above, the lack of new ground to cover leads to a claustrophobia manifested in endless arguments about what should and shouldn’t be on the wiki and non-stop edit wars.
This happens in other communities too, by the way. At some point in a political forum if there isn’t new blood people feel like they are just rehashing the same conversation over and over.
And so you get what I call Colony Collapse. One day you reach a tipping point, and suddenly the only people left on your site are the people you actually considered banning at the beginning. (This is what has happened to my old political community). For a while these people maintain the site, but eventually they too get bored and leave, and the site falls into disrepair. It starts to rot.
You can see this at my old site Blue Hampshire, which has reached the final phase of collapse, and now consists primarily of syndicated political press releases and occassional comments about how moronic other commenters are. There’s 8 years of beautiful posts in that site, almost 15,000 posts. Many contain wonderful explanations on how New Hampshire government works, personal reminiscences of New Hampshire political history. There are comments on that site with more political wisdom then you’ll find in a year of the Washington Post (there’s over 100,000 comments).
And it’s all rotting.
It’s heartbreaking, and after you’ve been through it once you get into a feed-the-beast mentality about all future communities. To paraphrase the line from Annie Hall: Online communities are like sharks; they have to keep moving forward or they die. And so, as community leader, you take on the exhausting role of the shark, pushing the site forward, always watching for the dreaded inflection point which presages the site’s collapse. Because once that happens, it’s all over.
A Different View: Wiki Sites as Bounded Conversations
That’s a long digression. But back to federated wiki.
Here’s a possible vision for federated wiki sites that you, the user make.
You’ll make federated sites for conversations on things, the way we are having this happening on Teaching Machines. And during the event, you’ll build it out. And then, at a predetermined point, you’ll call your personal version of that site done and abandon it.
In other words, we bake heat death into the plan. We accept our mortality.
That’s great, but now comes the question — how does the material get maintained? How do we recurse over it?
The answer we’re coming up with is that it gets pulled into new sites belonging to new conversations.
We have an example of this right now. We tried an experiment a while back that was a bit of a flop — a wiki called the “Hidden History of Online Learning”. It’s got about a hundred pages of this variety in it:
The site flared up into activity for a couple of weeks, then sputtered and died. By all normal metrics it’s a failure, doomed to bit-rot and link-rot, a slow descent into a GeoCities-like hell.
But in this case everyone involved has their copy of the site. As they participate in the Teaching Machines happening, they can pull stuff over into the new teaching machines wiki they have made. They’ll spruce it up, check the links, and maybe even improve it a little.
Ward and I were talking about this, and he said it reminded him of an earlier time when he was first working with Smalltalk. Objects would get better the more systems they were used in, because each time they were reused they would get refactored, optimized, simplified, extended. He and his fellow coders even had a name for this: “Reverse Bit-rot”.
It shouldn’t happen. It defies the entropy we see in all those pretty graphs up-page. But it happened.
Maybe this can happen here too. Sites can end, like conversations end. But we reach back into those previous conversations and say — you know, we were talking about that a couple months ago, let’s pull this thing in. And that thing gets another spin.
Maybe this doesn’t sound radical to you. Maybe it sounds ordinary. But I am sure every former online community manager out there understands just how radical this idea is. It’s equivalent to the soft-forking solution that was adopted Live Journal that became *the* solution for the non-collaborative communities that followed. But here it is applied to collaborative tasks.
In this world, sites like Blue Hampshire die — and maybe even die quicker, more humane deaths. In fact, maybe we say, hey, here’s a site that is going to exist just for the 2008 election. Two years later you spin up a site on 2010. Four years later on 2012. Some material from 2008 flows in. Some doesn’t. If the material is good enough to live on, there is no failure, no Colony Collapse.
Each wiki site is the product of a bounded conversation, expected to die, but also expected to be raided for the next conversation.
Kind of nice, right?
13 thoughts on “Rethinking Wiki Lifecycle: Sites as Bounded Conversations”
Nonononno u broke my heart
Then i thought, hang on, it’s like a PhD thesis or book that u eventually publish and move on. Sorry for the print and static examples.. But it makes sense, sort of, now that i think of it.
But gosh i did not know i felt so strongly about fedwiki
Well, there’s different ways to think of it. People can certainly keep the one running log and use it for everything.. I plan to keep at least one “daily” stream, but then share out to these other things.
Another way to think of this is “sites as hashtags” or categories.
Right, so how would this look like? I would keep maha.uk.fedwikihappening.net (need advice on moving to self-hosted btw) and i’d have a “happening folks” for our December experiment, and a “Teaching Machines folks” on my homepage, which takes me to a diff group of ppl? So we’d have to opt in for each one of these happenings?
there is a very simple procedure described here to setting up self-hosting – http://mura-fedwiki.englishup.me/installing-on-digital-ocean.html
Yes, exactly. Each would be opt-in. But I think we might also have an “all-wiki” conversation club page, so you would be able to get a global view as well.
Re: self-hosting, this page does a good job explaining the steps:
If you want to migrate your files en masse there are some additional steps (downloading files, and uploading and extracting them. Personally, until we get drag and drop archive uploads its faster just to fork em all over.
Maybe I’ve been hanging around an anthropologist and a dissertation writer too much this week or perhaps I’m just feeling a bit Polly Anna today, but I have to think we haven’t seen the end of sites that have died. Perhaps they’re dead in terms new knowledge production, but we don’t know how they may be used in the future. What boon of information for some future researcher some of these sites may be–we just don’t know (fist bump, Polly Anna!). For instance, my dissertation writer friend is looking at how political ideologies are passed down or ignored generationally, and all of his primary documents are letters, novels, newspaper reels, and books on or related to the topic. It’s been a tremendous amount of information to sort through–and that’s the useful practice of doctoral work. Since his work is about American politics, I can’t help but wonder how his writing would be so different if he had a Blue New Hampshire-type site to refer to or search in as he was formulating his own questions. How the comments may have redirected him. How the dead links could send him another search. Dunno.
My anthropology friend thinks the fedwiki has so much potential in terms of looking at how humans form ideas– My stories of working with Maha and Frances blew her mind. So my inner Polly Anna wants to wait and see before I write the epitaph for dead sites. But then again, I’ve never built something that big and watched it die slowly as you’ve described. FWIW, I’d like to keep my journaling connected to whatever comes next–everything I do with digital collaboration feels like the colony collapse you’ve described. Everything I have in the fedwiki feels like a productive little hive where I’m both the Queen Bee and a worker bee. But you know, I get it. Sometimes you have to build a new hive someplace else.
So as somebody who participated in The Happening, would I be able to bunch all of my current work in the fedwiki into [[The Happening]] page? Then I could start new with [[Teaching Machines]] if that’s possible–just like I’ve done with December and January journals. I’m all for whatever you and Ward are planning, really, but I just had to throw it out there that I’d like to keep the massive mess I’ve created connected to “the new conversation” as you’ve put it. Maybe I’ve missed the communique about this, so I wanted to ask: Will I be able to keep what I have from The Happening and attach it to The Lyceum? If so, can it look like what I’ve described?
Alyson — (and maybe Maha too?) I think the import of much of this is probably lost if you’ve never run a large (or moderately sized) online community. When I say sites are “dead” what I mean to do is challenge our notion that an inactive site is is a bad thing.
A persistent problem with online communities is the mission paradox — when a site more or less meets its mission, it disintegrates. That’s perceived by most to be a *bad* thing, a failure.
But the reason *why* it’s seen as a failure is actually that the materials in the site are locked into the site. That’s why we work so hard to keep communities past their prime running, because with the community dies the content.
Here that’s not the case. When sites die, the pages in the site don’t die, they migrate forward into new contexts. The paragraphs move on. The only thing that dies is the site’s interaction with that conversation. Again, I think maybe the import of this is lost if you’ve never organized online communities, but it’s a massive thing.
I see what you mean now, thanks for the clarification. All I could think of is all of the lost work that I have in LMSs of the past. Once we signed a new contact to the next LMS, well, all that work is dead. That’s my tiny way of relating to the “massive thing” you’ve described. That’s why the fedwiki is so frickin’ wicked cool with potential.
Doesn’t each *member* make this choice, since each member ends up with copies/forks of the pages he cares about?
So even if you spin up/off/out a new site, that’s *your* site, and if I take forks from both sites into my single site…?
I’m late to the conversation here, but I think Mike is missing something here. The “decaying” hidden history of online learning farm was created around the project. As the project wound down, activity waned. Contrast that pattern with an hypothetical one in which many users have permanent personal fedwiki sites.
Mike creates an HHOL page on his site and perhaps invites people to fork and contribute, which they do on their own sites. The hhol site doesn’t decay because there isn’t an hhol site. Instead, HHOL is a collection of pages and links that live in shared fedwiki space.. My take was that HHOL was only a farm/site in the first place because Mike was the only participant in the project with his own fedwiki at the time
Nearly all of human activities follow a live-it-up/slowly give-it-up cycle that is tied to each generations peer-acceptance. As one generation rises, another falls. The earlier generation is involved in activities, each of which is either is admired and emulated, thus extending that activity, or if not admired, replaced by another invented activity. It seems each generation strives to be different in some ways from what has been and wants to be different. This leads to new accepted norms that tend towards a sameness in that differenct. But in essence the value of a given activity, and thus its lifetime lies in what I call its basis, which I call its U3 properties. see ( http://c2.com/cgi/wiki?UsefulUsableUsed )
This also fits the notion that peers “want to different but/and to be the same”