Category: development

Why personalization is hard (and why you need to aggregate.)

zite

 

Since it launched, Zite has been one of my favorite apps. Keying off my Twitter stream the product does an absolutely amazing job of finding stories I am 1) interested in; 2) had not seen already; 3) from sources I was unlikely to stumble across.

That sort of personalization – saving time by accurately filtering information –  is of huge value to readers. If you can reduce the number of clicks (the currency of the web) needed to find stories of interest, I am likely to return. No surprise that most media organizations have recommended story lists on their websites.

But in many cases these lists are of minimal value. Browsing Salon.com just now – five of the suggested stories were more than four months old. And I was reading about the Affordable Health Care Act.

The challenge is, even the most prolific news sites (say the NYT) only publish 1000 – 1500 stories daily. Even over a week or a month, there is not enough variety to make an effective recommendation in every case. So, older or less relevant stories are surfaced. But as a pure aggregator Zite is not restricted to a single news source. A billion new pages are added to the world wide web daily. That is more than enough to recommend 50 – 100 highly relevant stories to me each day.

Doing some news math, the filtering value (or number of potential clicks saved) to the user is not comparable:

Zite: (one billion possible stories / 100 filtered stories) = 10,000,000 clicks saved.
News site: (one thousand possible stories / 10 filtered stories) = 100 clicks saved.

Yes, that is a farcical equation and the news site still returns a positive value — but not enough to be competitive.

For news publishers, there are three ways to improve our odds: 1) Improve the algorithms to eliminate outdated or irrelevant articles; 2) Reach outside your own CMS and aggregate stores from other sources; 3) Start focusing your metrics less on raw page views, and more on efficiency and value created for the reader ‘clicks saved.’ (See Cory Bergman’s ‘time saved’ take on this earlier this year.)

Yes, those aggregated links will need to be highly relevant (see improve the algorithm above) and they will lead visitors away from your site. But, all Zite does is send me away. And yet I return every day.

Using Google Forms on a smartphone

I did a fair amount of Google-ing on this looking for a simple answer – without much luck.

So, in case this comes in handy here is a simple-ish way to use Google Spreadsheet forms on an iPhone for data entry in the field.

My current project is an ethnographic survey of news habits among commuters – with a focus on mobile devices. I have been gathering anecdotal info for a while on the commuter rail and T.

For class, I needed to formalize the data a bit, and wanted to use Google Spreadsheets. There may be better survey solutions out there that are mobile-friendly, but I have a lot of other stuff in Google Docs, so there you go.

Creating a form to gather Google Spreadsheet data is dead simple. And (previously unknown to me) you can break a survey form into multiple pages and contextually customize questions. That is GREAT for mobile use as it reduces the amount of content on a single page, and eliminates non-applicable queries on-the-fly.

Unfortunately Google Forms does not include any mobile-specific or responsive themes. It does work on a smartphone, but only with some significant pinch-zooming. This is especially a pain when using tiny radio buttons or check boxes.

The solution I arrived at is pretty simple, especially if you have access to a web server.

1. Set the theme for your form to “Plain” in Google Docs.

2. Limit your survey pages to one question each with 4 – 5 answers – eliminating the need to scroll to find the continue/submit button.

3. Use page breaks, and multiple choice options (which allow a redirect to specific pages based on prior answers) to customize the order and questions offered.

4. Embed the survey code from Google into a web page you control. Doing this on your own server is preferable as you want a completely blank page aside from the embed code.

Screengrab of a simple mobile survey page5. Drop a meta tag in the <head> to set a viewport. Something like: <meta name=”viewport” content=”width=320, initial-scale=.9″> I have not touched HTML much recently but viewport is not tough to understand and works on Android as well.

6. Set the iFrame width to 320 (or etc for your phone) and the trim the height to make it as short as possible, without losing the bottom of any page.

7. Test and adjust the viewport width or scale as appropriate.

8. Keep in mind this approach is not the most elegant use of a viewport, but my goal was iPhone specific so I was satisfied to hard-code these settings into the page.

9. Once you finish tweaking, you should end up with a decent looking mobile form, as pictured here, without the need to pinch-zoom every page constantly.

10. Suggest to Google that mobile-friendly themes would be great.

Seven mobile trends to watch in 2012

Arguing Apple vs Android and apps vs web is fun, but so 2011. So, thinking about 2012, a handful of mobile trends are worth tracking:

  1. Transactions/Authentication (NFC , Square etc.)
  2. External sensors and connected devices (Bluetooth 4.0/Internet of Things)
  3. Voice (Siri vs Google)
  4. Presence (Moving beyond check-ins)
  5. Home Hub (Airplay, HDMI outputs, home controls)
  6. Connected cars (3G-enabled, streaming Internet replacing AM/FM etc.)
  7. 4G (Speed changes behavior)

I am collecting links on these and other mobile topics on Delicious.

Live from 39,000 feet

What is it about new toys?

My flight back from Las Vegas this afternoon (Southwest 1159) was on one of the airline’s few planes that have WiFi hotspots. Apparently they are still testing out the system and it was free.

So – having a new iPhone, 5.5 hours of flight time and free Wifi – of course I had to check it out. First up was Qik:

http://qik.com/video/1880919

Not too thrilling – but my son did get to watch the live video for a while which was fun.

I also tested out Skype- which worked like a charm on my end – though Annette could not hear much over the engine noise. Need a noise-canceling mic apparently.

The VPN connection into my office network was no problem. More fun was the SimplifyMedia app for the iPhone – which let me stream my home iTunes library for awhile – till I decided to save my battery I can imagine the Slingbox app would be pretty cool as well.

In the end I spent a fair amount of time watching our flight progress on FlightExplorer.com mostly just to watch the weather we were routing around.

And of course, I had time to write a blog post…

The Kindle/Google/distribution problem

An interesting quote from Jonathan Miller (once-upon-a-time my boss’ boss’ boss’ boss at AOL) talking about Kindle, the WSJ.com and the distribution problem in digital media:

I went from paying $14 to The Wall Street Journal to paying $10 to Amazon (for WSJ.com on the Kindle). Now the splits there, and I think this is relatively well known, are very, very much in favor of Amazon. So I became very much less valuable to The Wall Street Journal. That’s part one. Part two is they don’t know I exist. I went from being someone who’s their subscriber to being someone who is an Amazon subscriber, which The Wall Street Journal has no visibility back to and cannot manage that customer relationship. . . . So they’ve lost both the customer management and, trust me, the lion’s share of the economics.

So newspapers are mad at Google for creating an efficient distribution system that drives traffic back to them, but the same publishers are rushing to Amazon to give them 70% of the subscription revenue to get onto Kindle?