Walt Crawford has spent a fair amount of time this weekend trying to figure out why Bloglines isn’t picking up his posts. And he isn’t the only one.

Flickr had its own problem this weekend. On Saturday morning, I was looking at my contacts’ photos on Flickr, and I noticed that Laura Crossett had what looked like some interesting architectural photos up. So I clicked on a thumbnail of a lighting fixture, but got back what looked like a lo-res film still of a guy in a parking lot. Back in my photostream, a thumbnail photo of my kids was replaced with a busty anime character, though when I clicked on that, I got the correct image on the photo page. “WTF?” I thought.

Apparently a lot of people were also thinking “wtf?” and saying so in the Flickr forums. I like this quote from a forum post by Flickr user Dr. Keats: “55 out of my 860 [photos in his photostream] have got the dodgy photostream image – some pixilated, most not. Two are porn, the rest aren’t. Some are, in fact, very good…”

While many people thought the site had been hacked, it turns out the the official explanation is a bad caching server.

[Side note: the fact that so many people found porn/adult images in their photostream is an interesting indicator of how much of Flickr must be "adult" images (since we have to assume the images were being re-assigned randomly, yet many people complained of porn). There must be a whole 'nother world of private/NIPSA photos beneath the surface of Flickr.]

This got me thinking a bit about the Five Weeks to a Social Library class, where many of the participants are just starting to explore sites like Bloglines and Flickr, and some have expressed some understandable anxieties about it all.

The Five Weeks organizers have already had to tell participants that Bloglines may be buggy, which, I suppose, isn’t really all that big a deal; your choice of aggregator normally just affects you, though you might find your blog’s readership suffering if too many of your readers use Bloglines. But when Michael Porter and I talk about Flickr in week four of the class, I’m not really looking forward to saying “Flickr is a great place to store the photos for your blog; unfortunately, there is a small possibility that one day you will wake up to see photos of naked dudes on your ‘Kids Club’ blog. Welcome to perpetual beta.” (Or gamma. Whatever.) I don’t think that “Flickr is having a massage” (or “your library is having a massage“) is gonna cut it.

And the fact is, it’s Flickr and Bloglines this week, but next week it could be Blogger and Typepad (both of which have had their rough spots over the years) or FeedBurner and PBWiki (neither of which have ever given me trouble, but if they did, I would feel hung out to dry).

I’m not sure what the lesson is here. It would be nice to think that we could do all this ourselves, or minimize dependencies as much as possible, but most of us can’t write this software or even host it ourselves, and there may be features that we want to take advantage of (such as the subscriber count in Bloglines or the social features of Flickr) that simply couldn’t be duplicated.

I guess the questions to ask are “what could go wrong?” (though I don’t think I would have thought of the Flickr-photo-roulette thing until it actually happened yesterday); “how risk-averse am I?” (or “… is my institution?”); and “what is my exit strategy?” (e.g., do you keep your own blog/wiki/photo backups in a format suitable for importing to another application if your current application becomes unusable?).

We already know from our experience with library catalogs that putting too much faith in vendors can be a mistake. Are we making the same mistake with social software?