DISQUS

Fred, does HeyZap have any barriers to entry, or means of locking in early mover advantage? i.e.: what's to prevent Yahoo Games (http://games.yahoo.com/free-games) from widgetizing their inventory and steamrolling smaller players? It strikes me that HeyZap is pretty exposed.

Fred, this calls to mind this great article from Fast Company:

http://www.fastcompany.com/blog/gadi-amit/new-d...

In my work, I'm constantly struggling to both institutionalize capital-i Innovation, and still leave room for intuition to flourish - i.e.: leaps of faith.

Not to overhype the connection to the finance industry, but I wonder what a company like Angoss would have to say about identifying spam?

http://www.angoss.com/

That's essentially what they do for financial risk; identifying claim spam for insurance companies, etc.

Dave, thanks for sharing your thoughts. I don't know enough about the skillsets of each of the team members you noted -- the modeler may very well have things covered in regards to the statistical rigor of source-blending.

Whatever the case - best of luck!

-R

Fred, Dave - how do you account for adverse selection in using Twitter derived data in a decision-making setting? it strikes me that that is the larger question.

Set top boxes are passive, Twitter participation is active - that is to say, Twitter metrics will select for certain user classes. Identifying those classes and extrapolating them into larger population trends (i.e.: blending twitter metrics with traditional metrics via some credibility system) will be the challenge that takes something like this from "curiosity" to "core" and "actionable."

I see a lot of parallels to the statistics and mathematics applied in the insurance industry (experience analysis). If you're working along these lines, you may wish to consider retaining a consulting actuary.

The singularity is nigh!

// re-reading Accelerando / Charlie Stross in Stanza. Its CC licensed and free via FeedBooks.

If you want to have some UI fun with GMail, go into Labs and enable "Mouse Gestures:"

"Use your mouse to navigate with gestures. Hold right-click and move the mouse left to go to a previous conversation, move it right to go to the next conversation, and move up to go back to the inbox view. Works best on Windows. [but works fine elsewhere too...]"

If you're into mouse-less usage, however, this may not be for you.

Huh. Remember Uncov.com and their startup "persai?" IIRC, persai was supposed to be a personalized, learning aggregator. I popped by to see what state it was in:

http://persai.com

Now redirects to "pressflip.com" which appears to be a really thin "persistent search" service over google.

too bad. After all of the vitriol of uncov, I'd hope for something at least interesting from the team.

Hi Robert - I noticed yesterday that you subscribed to one of the TechWatching Twitter feeds (http://twitter.com/techwatching) - thanks, and I hope you find it informative and timely. That feed is the output of http://techwatching.com

I wanted to share a bit of background on TechWatching (TW) with you, tying into the theme of innovative memetracking.


(1) Like its progenitors (TechMeme, etc.), TW reads feeds and links together stories based on link behaviour, keyword analysis, and human editorializing.

Using Twitter, TW takes things a step further, however: Twitter provides attention data that feeds back into my algorithm. As each story is presented in concise, identical fashion on Twitter, I view twitter click-throughs as great organic indicators of interest - i.e.: clicks aren't driven by page placement, font size, or what have you.

The twitter attention data seems to be driving some interesting results, generating a page that covers many of the same things as Techmeme, but that also surfaces some really different stories.



(2) TW adds wiki features to the mix. Click through to a story cluster:

http://techwatching.com/cluster/133079

That page is broken down into the "cluster" at the top, which includes stories that either explicitly linked to one another, or have been "editorialized" together. Below are "related" stories, collected based on simple keywording, that anyone can add to the cluster or remove from the page.

This is all pretty experimental; I don't know what kind of success or abuse a "wikified" memetracker will experience. But: I'm watching it happen, and iterating the features rapidly.


FWIW: TechFuga looks like its suffering from many of the issues that I've worked through or am in the process of working-through on TW - false story correlations (i.e.: everything "blackberry" being lumped together even if the stories are about wildly divergent blackberry-memes), as well as spurious keywords entering the mix (like the story clump apparently based on the word "day" http://www.techfuga.com/23122008/two-days-until...). Good luck to them. They have a long way to go, and turning to AllTop-style meta-analysis of other memetrackers isn't going to help.


Anyway: I hope you find TW useful; if you've got a moment, it would be great to get your feedback.

Regards,
--Rod Edwards

Cheers, Fred - I appreciate the positive feedback greatly.

FYI, here's one of yours that just bubbled up:

http://techwatching.com/cluster/85227

Its not on purpose, no. There's a few factors that contribute to that.

1: The gadget blogs interlink, a lot - i.e.: engadget, gizmodo, crunchgear - if one covers a story, they all do. That interlinking pops those blogs up a notch in the algorithm to begin with.

2: They tend to get a lot of clicks - people appear to be very interested in gadget news both mainstream and esoteric.

3: Right now my seed list of feeds is probably too gadget weighted. Some of the off-shoots in particular (Engadget Mobile) should be downgraded in the algorithm. That's a symptom of the metrics that I use to choose which feeds to add to the seed list; Engadget Mobile, for instance, looks like a good seed candidate based on the number of inbound links it receives, but a large portion of those links are from other Engadget properties.

4: The gadget weighting also seems to be a Monday-morning phenomenon due to the relative lack of business news over the weekend; as more business/tech news develops in the work week, the prominence of gadgets will decrease.

So - to answer your question, its not on purpose - but the algorithm is performing as designed, based on the inputs I've given it, and the data it collects.

Here's something for you: I run a memetracker as well, its a weekend project for me that I've been fooling around with for a year or two. About a month ago, I started using the Twitter API to post items from my tracker to twitter, and was pleased to see some take up.

Now, though, I've taken it a step further: Twitter provides attention data that feeds back into my algorithm. Because each story is presented identically on Twitter, I view twitter click-throughs as great organic indicators of interest - i.e.: clicks aren't driven by page placement, font size or what have you. The twitter attention data seems to be driving some interesting results, generating a page that covers many of the same things as Techmeme, but that also surfaces some really different things.

Have a look at the feeds: http://twitter.com/techwatching and http://twitter.com/techwatching_cl

For What? is exactly what I thought when I read the press release/blog entry from Adelson too. Their answer to their inability to break out of their community imposed niche in the states is to.... try again in different countries. Huh.

I suppose its the founder's imperative to always imagine that their property has the potential to be something more than another iteration of its current state, but in the case of Digg, I can't help but think that it is already what it will ever be: a center for a small, passionate community that will sustain a nicely sized small business, possibly getting acquired as an quick ad inventory grab.

You know, with stories like (and Microsoft investigating Siberian data centers), I've been thinking Winnipeg should be positioned as a data center hub - we have ridiculously cheap and green power (abundant hydroelectric), a good labour force to draw on, and a climate that makes cooling really cheap for a good portion of the year - talk about turning lemons into lemonade.

In fact, I emailed the mayor's office about this once a year or two back. But never heard anything. I figure a litte trade mission to Google with the mayor, a rep from Manitoba Hydro, and someone from the province could go a long way towards further diversifying the prarie economy.