Medill Ups Its Hacker Street Cred

Last week Medill’s WriteClick initiative announced some forward-looking projects that made me feel some pride in the progress my alma mater has made in the last few years.

The projects:

Tweedia, a widget that will allow readers to instantly see other relevant Twitter posts on news stories.

Machine Generated Sports Stories, a program that instantly “writes” stories about baseball games, drawing on data scraped from box scores and play-by-plays online.

EasyWriter, a Microsoft Word plug-in that will allow writers to make refined web searches within Word.

Twitter Publishing, an application that will allow media companies to send news links to interested Twitter users, based on what they post.

News Feed, an iPhone web application that utilizes time-driven news feeds to give time-constrained readers a more efficient way to read the news.

The Time article “Can Computer Nerds Save Journalism?” sums it up beautifully:

A cadre of newly minted media whiz kids, who mix high-tech savvy with hard-nosed reporting skills, are taking a closer look at ways in which 21st century code-crunching and old-fashioned reporting can not only coexist but also thrive. And the first batch of them has just emerged from Northwestern University’s Medill School of Journalism.

A hearty congratulations to @richgor and all of the other faculty and students who have contributed to these projects.

Moving Past Textbooks

I feel compelled to reblog Seth’s Textbook Rant:

As far as I can tell, assigning a textbook to your college class is academic malpractice.

They are expensive. $50 is the low end, $200 is more typical. A textbook author in Toronto made enough money from his calculus textbook to afford a $20 million house. This is absurd on its face. There’s no serious insight or leap in pedagogy involved in writing a standard textbook. That’s what makes it standard. It’s hard, but it shouldn’t make you a millionaire.

They don’t make change. Textbooks have very little narrative. They don’t take you from a place of ignorance to a place of insight. Instead, even the best marketing textbooks surround you with a fairly non-connected series of vocabulary words, oversimplified problems and random examples.

They’re out of date and don’t match the course. The 2009-2010 edition of the MKTG textbook, which is the hippest I could find, has no entries in the index for Google, Twitter, or even Permission Marketing.

They don’t sell the topic. Textbooks today are a lot more colorful and breezy than they used to be, but they are far from engaging or inspirational. No one puts down a textbook and says, “yes, this is what I want to do!”

They are incredibly impractical. Not just in terms of the lessons taught, but in terms of being a reference book for years down the road.

This post, Seth says, has gotten more instant e-mail replies than any of this previous posts. It struck me as a sort of coda, signaling that we’ve reached a consensus, and we’re ready to move on.

Rethinking Emergency Services

Yesterday afternoon, I called 911. I observed an emergency unfolding, and figured it would be easy and fast to report on it.

I was wrong. The local 911 operator put me on hold for more than five minutes. I’m not sure how long I would have been on hold for, since I hung up.

And while muzak wasn’t playing, it might as well have been. In an emergency, every minute counts, and there’s no excuses in 2009 for putting 911 callers on hold.

A Better 911 Call Procedure

As soon as my call connects to the emergency hotline, I should hear three words:

“What’s your emergency?”

Until the caller can be connected to a live human, the call will be recorded and analyzed - both based on sound and on semantics - to estimate the priority level of the call. If loud noises or the sounds of screaming or certain trigger words are recognized, then the priority level is increased.

These changes wouldn’t just help people in emergencies, but also observers who would like a quick, efficient way to do their civic duty.

One Protocol to Ping Them All

PubSubHubbub is a plug and play pubsub solution for App Engine deployment.

If you’re not even sure why you’d want pubsub over polling to begin with, the PubSubHubbub wiki has a page on Why Polling Sucks:

What’s wrong with feed updates every few minutes/hours/days?

It might be fine for reading your blogs (it’s obviously been fine for most people for years now), but it’s not fast enough for the decentralized applications of tomorrow that people want to build.

We and others are thinking about totally decentralized social networking, with real-time updates. Don’t assume any of this is about blogs just because we’re using the Atom format and Atom’s typically used for blog posts.

Atom feeds can be used to represent any topics, including private person-to-person topics such as direct messages or even real-time moves in a game.

Whether it’s for moves in a real-time game, livestreaming video, or remixing music, the common use cases of Atom and RSS and JSON and APIs and HTTP are quickly taking on a new light as polling gives way to pubsub.

In fact, if pubsub truly can fulfill the promise of 500ms or less latency, it would be a genuine Omega Point, as big a game changer as the RDBMS or the GUI.

Google Squared Is No Freebase-Killer

Now that the Google Squared site is public, it’s a little more clear that the site is comparable to a Freebase view.

Freebase has iteratively improved its View interface for a couple of years and super-classed it with the Bases feature, so it is generally more full-featured to Google Squared, although the responsive minimalism of Google Squared could make it a better interface for mobile devices.

And yet, despite Google’s resources, Freebase has a major advantage in that it is solely focused on interfaces for its data, and it comes up with all kinds of clever ways of improving the data. A perfect example would be the announcement today on its blog:

New on our topic page: help us type untyped topics

The announcement is that Freebase is now using human intelligence to confirm estimations about the types of untyped topics, using Wikipedia Categories. This is something we’ve been discussing with the Freebase Team for a while, but it’s a total surprise that this is now a live feature on thousands of Freebase topic pages.

Already having done some work using Wikipedia categories to supplement Freebase topic data, I’m going to be very eager to follow this confirmation feature, and see if there is a way to ombine it with the new eMQL adapters.

Update: The source code for the Typewriter app is available, but it lacks documentation on the mechanics behind the estimations.