Sunday, September 13, 2009

Twitter search: ephemera for ordinary users so it can be something more for business intelligence?

ephemera: items designed to be useful or important for only a short time, esp. pamphlets, notices, tickets, etc.

I have been thinking for a while (and I’m not alone) that Twitter search has been consciously crippled for a good reason. That is, it’s good for Twitter, still publicly in search of a business model, but obviously trying models on for size. It’s not so good for users.

Here’s an example. I was interested in the inflated attendance counts for the 9/12 events and did a search on “ABC million” for Saturday and Sunday. (A FreedomWorks speaker pulled the figure of 2 million from someair, falsely or erroneously attributed it to ABC News, and word spread across Twitter like a swarm of gnats.)

So, the search proves useful, for the moment, and though the moment is what drives Twitter, you might want a record of the reaction, and your desire would be thwarted. Twitter search results go back only a week and a half at best and developers reserve the right to further limits based on traffic.

You could grab a feed of the search results, but it’s limited to the last 30 tweets. You can let the items pile up in your aggregator, but only starting at the point when you realize it’s a trend. History is important.

The ability to specify dates for a search is nice, but if it only goes back X days the utility is limited. Searching for a range of hours would be a great help, but it’s not available to users either. How would it help? Well, since Twitter is so “of the moment” the narrower the time span the more results you’ll get for a fuzzy query. Let’s say Rachel Maddow says something provocative and I’m not ambitious enough to construct a complicated query of a string of “Or’s.” If I could specify that I only want tweets tweeted from 9 until 10 p.m., most every tweet containing “Rachel” would be in reference to the show, and I’d gain the added benefit of seeing mostly viewers who are tweeting while watching.

So why don’t users have these superpowers? It could be because Twitter needs to curb features to keep from crashing, but it might be that we aren’t given the value-add precisely because it is so valuable.

Can’t you see the charts in the marketing reports? “Since the introduction of the X campaign, mentions of Product Y are up 23% over the previous month and up 67% over this month last year.” Historical trends are gold. Twitter has to be selling the data that could produce them, or they’re holding it back from us while they think about how they could sell it.

Am I an anti-business socialist? Not really, but this sort of behavior on Twitter’s part does bring to mind the traditional labor phrase, “on the backs of the workers.” Twitter users and the words we type are the Twitter product. We’re stakeholders, so if our aggregated facts, sentiments and opinions are on the block, we should get something back—not money, just utility.

Sunday, October 26, 2008

Trying a few things

New blog. I’ve been changing blogs like John McCain changes his strategy, but I haven’t worried about preserving my brand since Google shrank the pagerank on my OPML blog from 6 to 0. (Long story.) I figured, like Sarah Palin, what do I have to lose. So, when the OPML Editor didn’t play well with my new desktop running Vista, I moved from the OPML blog to the blog at learnandteachonline.com that I abandoned three years ago. But it’s made with Drupal, which I just never took to, so I never went deep into it. Every time I upgraded or wanted to change something, answers on the support forums were circular and haughty. I dreaded working on it.

So I’m making this with Expression Engine (EE), the LAMP CMS I use at work. It feels like an old shoe. I’ll just start fresh, and not worry about trying to migrate old content. Lijit search is in the sidebar and is supposed to include the old blogs, though at the moment it doesn’t seem to be finding everything.

Trying Friendfeed for comments. I’ve been following the developments at Friendfeed on the Gillmor Gang and NewsGang Live. On a recent show someone, probably Steve, mentioned using the service for blog comments, and it reminded me I’d thought of trying an FF room for that purpose shortly after the rooms concept was introduced. 

You can see the widget for the almost real-time Amylooo room in the sidebar here. I’ve also sort of integrated it traveling the other way. The comments link at the bottom of each post is a link to the Friendfeed entry. It’s not completely automatic; I have to publish the post, then go to Friendfeed to fetch the link, and enter it in a custom field in the EE blog post form.

Entry link naming at Friendfeed is quite complicated, probably because each comment in each thread has its own URI, and maybe for some security purpose?  That’s a good thing, but it’s too random, not predictable enough, to integrate it the way I can with some services. It would be nice if each entry in my room started with the same string so it could be appended with the post title here to save the manual work. For example, for a sidebar item in an email newsletter I do at work in EE, I can pop up a page for a poll at Survey Monkey that only requires an ID number added to an address that always starts out the same. 

Does your organization have a policy on cell phone use while driving?

Take the survey

Maybe there’s something in the API that would facilitate that. I haven’t studied it.