I should probably stop talking about the federated aspects of Smallest Federated Wiki — as I mentioned before, whether federation works or not for any given use case is speculative. There is no way to come to real agreement about it since it relies on us making predictions about how people will act in a federated content environment, and no one really knows what that looks like at scale. In situations like this it’s probably best to just try to get people to use it rather than talk about models of use.

But Maha Bali makes some great points in the comments of the last post that I want to address. So I suppose I again risk the entire project by dragging people into its most contentious assertion 😉

Here’s Maha:

But then your point that wikipedia still has a place is a valid one: encyclopedias are not meant to provide the space for those diff perspectives, and wikipedia does have a discussion space for them, except not visible to the usual user.

I am just thinking of what it means to the “casual” web surfer (not the deeply versted researcher) to get confused by all the versions? There are already multiple websites for every web search we make, and we’re just usually using one search engine not a metacrawler or whatever.

First, let me agree with Maha that absolutely there is a place for Wikipedia in this equation. SFW is not meant to replace Wikipedia. But I want to address the “all the versions” critique.

Again, I think the issue here is we treat every piece of knowledge as if it were the Wikipedia page on the Civil War or Behaviorial Genetics. On those pages there’s a deep need for some consensus version that someone can read. But 99.999% of subjects are not like that. With a hat tip to Heavy Metal Umlaut, here are the 217 edits to Amelia Bedelia (more or less, speeding up the film may have dropped some edits).

Again, I can’t repeat enough the SFW is not a replacement for Wikipedia. But it’s really helpful to look at what *most* subjects look like, because our stress about “all the versions” is a classic case of over-engineering.

Of the 217 edits, most are vandalism and vandalism reversions. You’ll see that edits that flash by about Aspberger’s syndrome, “Booger Dystrophy” and the minimalistic “Dragons are cool”. You don’t see the war below the fold, but some of the edits are people adding fake Amelia Bedelia Movies below the book list (Sample title:  Amelia Bedelia in China (1897)). Other edits involve formating, italicization, small wording changes.

Filter out vandalism, style, and formatting edits, and there’s less than ten main components, really.

1. The main stub. What Amelia Bedelia is, who wrote it. Some level of plot summary
2. Some edits about other people involved with the work besides the author
3. A section that exists for a while on how Amelia Bedelia teaches kids about polysemy, etc.
4. The book list, with dates
5. A section that exists for a while on the vaudeville/irish connection
6. A section that appears and reappears about a statue of the author
7. The cover of the book
8. The link to the I Can Read Series
9. The Cameroon hoax

What happens if someone gets linked to an old version? Flipping through the edits, it’s hard to see the harm. Someone goes to a page to get a question answered, presumably. Even the first version of the page will answer most questions that people want to ask. The biggest risk is someone might miss the book list, which seems of substantial use to the casual reader.

Flipping through the edits, and omitting the book list addition, it’s often hard to see how a newer version is particularly better than an older version.  In many cases, for many users, the newer version is worse or less useful (not even taking the Cameroon hoax into account).

This is not the story we tell ourselves about Wikipedia. But that’s because we know where Wikipedia really succeeds — it does an amzing job on the big pages, with a large vibrant community who engage with the page over a period of years, and make sure that items don’t fall between the cracks.

But if there’s one thing I’ve learned recently, again and again, it’s that much of the web is engineered for a scale that it doesn’t operate at. And to a certain extent that is true about Wikipedia. There’s no other Wikipedia, because Wikipedia works best at scale. It tends to not work as well with Amelia Bedelia-sized things.

So what the harm? As Maha mentions,  it’s all in the edits, right? If you want to dig? Not really. I sped the film up: it took me 12 minutes just to click through all the edits, primarily due to rampant vandalism. That doesn’t include reading them, diffing them, or anything of that nature. No one’s going to do that, short of writing a book on a subject.

And (primarly because of the vandalism issue) past versions are not exposed to Google.

So when that fact is gone, it’s gone. As an author, you give Wikipedia stewardship of your work, but you do in fact also give them the ability to erase it from the web.

Let’s replay history with SFW in the picture.

I’m looking at Amelia Bedelia one day, and I realize — wow, this looks a lot like anti-Irish vaudeville.

I look for an SFW page on Amelia Bedelia, I fork it to my site, and add my insight. If people like that insight they pull it back. The people who care most about scholarship around children’s books fork it back.

A person on Wikipedia finds my idea in a Google search, puts it on Wikipedia, it gets deleted eventually.

The vandalisms, including the Cameroon vandlaism, never happen, because in SFW the only house you can vandalize is your own.

You have a question about Amelia Bedelia. You do a Google search. You get Wikipedia, but you also get links to a number of SFW, ranked by a reputation algorithm. If Wikipedia answers your question, great. If not, you have these links.

You click into a SFW page on Amelia. Which one? From Google, the highest ranked one, which is probably the one most linked to, or the one one the site with other articles of high reputation on Children’s Literature, or whatever. You read that one.

If that doesn’t answer your question, you look up top and see there is a list of more newly updated ones. You pick the most recent one. In each case, you always have access to an author page which tells you why the curator of this particular page should be trusted.

At all stages of this, there’s never too many versions. There’s only what you are looking at, and additional versions if you need them.

The “too many versions” problem reminds me a lot of when AltaVista came out as a search engine, and suddenly people were very stressed about at all the search results for a term. There’s this stress that information is out there that we are not dealing with, but the reality is the way we deal with it is decide at each point whether it’s worth reading more.

My wife Nicole, for instance, would love to see lesson plans on SFW. She could jump to one on Van Gogh.

How would she know that it’s the “best” one? She wouldn’t. Largely because there isn’t a best one. But having the different versions visible allows her to quickly sort through them to get what she needs. Over time, the best one ends up being the one that everybody links to externally (e.g. tweeting “best van gogh lesson plan ever!”) or ends up forking to their own site.

Anyway, I should get off this topic — as I mentioned, I’m trying to steer away from the federated aspect of SFW because it’s the one point of contention in a product that has many other things going for it. But the answer, roughly, to how we deal with multiple versions is that we let networks and network algorithms sort them, versus use a “last edit/best edit” system.  For large complex articles with active communities this won’t be the best way to deal with things — last-edit/best-edit works great. But for most things we do it will in fact be better, and in fact help Wikipedia to be better as well.