Blog levels

Software programs often have features for describing what they're doing. The usual term is log. Log messages are generally categorized by what part of the program is writing them. Usually there will also be a time and date. One of the key parts of a logging system is the idea of a log level. That's an indicator of how important the message is. Common log levels might include DEBUG, INFO, WARNING, ERROR, and FATAL. You can tell the program to tell you everything, or to just tell you stuff that's really important. You set a minimum threshold of importance, and then it just reports stuff that's at least that important.

I'm starting to think I want blog levels. Maybe the levels I'd have would be TRIVIAL, SMALL, MEDIUM, BIG, and HUGE. This post would probably be a SMALL. A mildly amusing picture, like the Amazon oops, that would be TRIVIAL. Changing jobs? BIG. New baby? HUGE. And so forth.

You are apparently interested in me. I don't question that; I'm kind of afraid of the answer. Nevertheless, you're probably not interested in every stray thought that crosses my head. You probably don't want to run this weblog on TRIVIAL. I've already got categories. Those aren't quite right, though. You can select the posts in a category, or you can see everything. But there's no way to say "only show things more important than MEDIUM." It's nothing or one or all. So yeah. Blog levels. I want them.

Search reruns

Every search on every site should have a little button you can press that says "Don't show me this again." If only I knew of sites with searches...

Also, while I'm on the subject... If you can refine a search by, say, geographical region, you should be able to pick multiple ones, not just one. Maybe you want to look at Austria and Switzerland, but not all of Europe. There's a middle ground between one and all. Actually, that particular example isn't all that useful. Amazon, though, could go to town with it. Newegg already does.

Restricted access

Linkedin annoys me every time I go to them because they're asking for my Google/Yahoo/Facebook/Whatever logins. Uh, no. I accept that some people find this kind of thing useful. But giving up my username and password? SRSLY? Uh uh.

There's a simple thing all these web sites can do. They can add restricted roles support. I should be able to log into Hotmail, create a restricted role called "Linkedin," give that role permission to look at my address book but not my mail, and then get a generated username/password combination to give to Linkedin. If Linkedin starts to annoy in some other way, say, by plaxo-ing everyone I know, then I can turn off their access. That's the basic level.

The even better level would be to be able to specify which information in which parts of the app the restricted role can see. Not just "Linkedin can see address book," but "Linkedin can see the first names and last names but not the email addresses." That way you can search for people you know en masse but avoid giving Linkedin the ability to spam.

Just thought of a social networking site for pigs: oinkedin(.com)

Freebase Parallax

Perhaps Freebase is something that can beat Wikipedia. Check out this video about their tool "Parallax." Parallax enables a model of browsing that automates what I have found myself doing manually. I have not yet tried it, so I can't say how it works, but they've certainly grasped a novel form of information retrieval. I'm not aware of anyone else doing something similar.

GimmeMyData.com

That's one possible name for the Web 2.0 backup service. The .COM domain is available.

Web 2.0^2.0

One worrisome trend with the shift to web-based applications and services is that your data no longer live under your control. Companies have outages, cancel products, go out of business, etc. With local data, you can back up to physical media or use an online service. Data important to you can disappear, and you can do nothing about it. This is a problem, which means it's also an opportunity.

Any Web 2.0 startup worth using gives you access to your data. GMail supports POP and IMAP, Blogger and Facebook have APIs, and nearly everyone has some kind of RSS feed. It's too much to expect the average person to use those protocols directly. What we need is a product to do it for us, something that knows how to talk to Flickr or Tumblr or whatever and get a copy of our data somewhere safe.

This product would take one (or both) of two forms. One would be a web-based service. That's a gimme for a Web 2.0 offering. You pay them $5/month, they suck down and store your data. Naturally, they themselves would need a 2-way API. They'd have plugins for all the sites their customers use. You'd be able to view your emails or tweets or whatever on the site to make sure they're there. You'd also be able to download all that data in a single blob that you could back up yourself locally. Perhaps there would be two levels of service, one where they store your data, and the other where they merely provide a single point of access. They would also provide some way to reconstruct an account from their backups in case the original service had a catastrophic failure.

The other possible form would be as a desktop application. After all, if the goal is to protect you from failures in web-based services, a web-based service might seem beside the point. The desktop application would do exactly what the web-based service did, except the data would be stored locally. What you did from there would be your problem. You pays your moneys, you downloads your softwares. I can see a good case for either form, or even both.

If this is such a good idea, why don't I do it? Simply put, the risk is too great at this time in my life. I can't take 6 months off unpaid to work on something like this. Mortgage, kids, insurance... It's too much. However, I do know a few people (hint hint) who are ideally placed for this. I'll even try to come up with a good name for it.

For Those about to Rock, We Salute Wikipedia

Heavy Metal Umlaut is my favorite Wikipedia article. None of the Wannapedias I mentioned has anything about it. Not on Conservapedia, Mahalo, Knol, Citizendium, or Squidoo. Final score? Wikipedia: awesome. Everyone else: l4me.

In denial about Wikipedia

5 years ago, you would have been completely justified in being skeptical about Wikipedia. About.com was an example of what was considered better. It was controlled. The authors and editors were qualified and authoritative. Wikipedia seemed like a ridiculous exercise in utopian naïveté.

7 years into the experiment, it's clear that Wikipedia is a success. Wiki works. What was healthy skepticism increasingly looks like denial. Mostly it's just verbal sniping, but there are a number of projects that attempt to "fix" Wikipedia by restricting authorship to appropriately qualified "authorities."

Citizendium requires real names and requires all articles be approved by their group of "experts." They're not different enough nor do they have sufficient critical mass to catch up. They don't appreciate what it is that has made Wikipedia such an unstoppable force.

Conservapedia aims to fix Wikipedia's "liberal bias." What they will do about reality's known liberal bias remains unclear. Their misguided aims doom them from the start; I imagine it will be no more successful than Air America Radio.

Mahalo tries to avoid authorship altogether. They provide no information directly. It's like they took the corresponding Wikipedia article and stripped out everything but the "References" section. Their compelling advantage is thus that they provide less information and require you to do more work. Good luck with that.

Finally, Google has entered the arena with Knol, which encourages anyone to post an article on anything, and let the search algorithm sort out who's best. I thought that approach was useless when it was called Squidoo; apparently Google likes the idea, as well as the idea of giving it a dumb name. Assuming people use it, you'll end up with dozens of different pages on a single subject, all incomplete in different ways. You also cannot make a small edit to an existing article; you have to build the whole thing yourself. Just like one feature does not make a business, a single correction does not make a useful Knol.

The one fly in the ointment for Wikipedia is Google's control of the dominant search engine. Thus far, they've been true to their "don't be evil" motto in their index; if their Wikipedia jealousy causes them to pervert their search results, people will just stop using Google and go directly to Wikipedia. I already do that for many of searches (Firefox's keyword searches are indispensable). And that doesn't even get into the shrieks of delight that will echo from Redmond (Microsoft) and Sunnyvale (Yahoo) as Google's competitors contemplate what the DoJ will do in response.

Every one of these alternatives throws the baby out with the bathwater. Wikipedia's genius is how it manages to be both centralized and decentralized at the same time. What's centralized is the collection of information, something Squidoo, Mahalo, and Knol cannot match. Besides the obvious advantage of having a one-stop shop, they ignore the importance of the snowball effect; to produce a better lens or Knol, you have to start from scratch. I would be surprised if any of my regular readers has never even once made an edit to Wikipedia.

The decentralized part of Wikipedia is of course the army of authors and editors, namely, the entire population of Internet users. It turns out that really matters. Conservapedia rejects anyone who doesn't share their set of biases (i.e., most of everyone) while Citizendium puts barriers in front of anyone who might want to contribute. If there's anything we've learned in 15 years of the Web, it's that everything you put between a user and an action, no matter how seemingly trivial, chips away at the number of people who will actually bother.

These sites may improve on Wikipedia in certain narrow ways, but those improvements come at such a cost that none of them will be able to defeat Wikipedia. Wikipedia certainly could stand a few improvements. Vandalism is generally contained, but still occurs frequently. The cabal of moderators occasionally gets unhinged. Wikipedia's markup has gotten increasingly complicated as the project has adopted more sophisticated conventions for formatting and organization, restricting what a casual editor can do. Those are all real problems that Wikipedia has yet to solve. Some day, something better will come along, but these guys ain't it.

A Huge, Ever-growing Brain That Rules From the Center of the Ultrawiki

I got a tip a while back from Rich to set up a wiki for personal/family use. Use the magic of Web 2.0 to run your life. I finally got around to doing it a little while back. My web host makes it really easy to set up MediaWiki, the software that powers the ever-more-awesome Wikipedia.

I've taken to using it more and more as I get used to it. Some stuff that used to go in my little now goes in there on a scratch page. I'm probably going to buy some glasses online soon, so I have a wiki page to track candidate frames. We have a page of home improvement projects, both completed (for reference at sale time) and TODO. I keep track of URLs that I discover at work that the censor proxy won't let me see, as well as URLs I discover at home that would be more easily read at home.

At some point in the future, we're going to move to a new house and buy a new car, so I've compiled a list of notes on both as I think of criteria and candidates. I've started tracking books I've read* on there as well. We'll probably start a page of movies to see, with indicators for movies we both want to see and ones just one of us wants to see. I won't link to it right now because some of the pages are private, and I haven't gotten around to fine-tuning the permissions settings.

That's just the list of things I've been putting in it for the last few months. It's handy because it's accessible from nearly anywhere, while also being easy to use and reliable; Dreamhost does daily backups for me. If you don't have a web host already, or don't have one that easily supports a wiki, or just don't want to bother, there are hosted solutions out there, like Google Sites (formerly JotSpot) and PB Wiki. You can see a more complete list of options at Wikipedia, in a nice little bit of circularity. Give it a shot; it'll change your life (if only a little bit).

* Still planning on posting here, but until I do, I need to record them somewhere.

PS - unembeddable reference

Paging lists of results

Reddit (I only read it for the articles) implemented their pager to remember where you last were, so you don't lose your place. I should explain that more.

Any site that presents lots of items, be they links, pictures, or goods, will break them up into pages. It's easier on them and easier on you (assuming the number of items per page is reasonable). Usually the way it works (roughly speaking) is that they sort by some value, like a price or a name, and then give you, say, the third group of 25.

That's all well and good for result sets that don't change much. Page numbers in a book make sense because the book doesn't change once it's printed. Web sites aren't books. Reddit gets new things constantly. If I'm on page 3 one afternoon, then the next morning when I click to page 4, I won't get what would have been yesterday's page 4. I might get page 3 again, the previous page 2, or a page of results that didn't even exist before.

What Reddit does instead is keep track of the last item on the page. When you click to the next page, instead of saying "grab the 25 links on page 4," it says "grab the 25 links after 't3_6ngfy'." That way you never lose your place. It's better for me, and quite likely better for them, too.

This isn't necessarily the right thing to do for everything that pages. Reddit to my knowledge never deletes links from their results (under normal circumstances). A retailer might do that relatively frequently, in which case this wouldn't work. It would be disastrous to refer to a section of a book by the quote that preceded it. This is not at all a brilliant technique. It's notable because it's different from the default. The Redditors realized that people use their site differently, and adapted accordingly.