Dispatches from Andyland

There were two videos that I came across recently that showed two aspects to current filmaking techniques. The first was
Stargate Studios Virtual Backlot Reel 2009 and the second as The making of Old Spice’s commercial: The Man Your Man Could Smell Like. What is interesting about them are the way they show two different aspects to movie production. The virtual backlot reel show how common simple video techniques like chroma-key (bluescreen) technology is. How common it is that the street scene, etc. that you see in  a TV show or movie has some amount of trickery done. The dissecting of the Old Spice commercial shows how much can be done with little or no trickery, and how much effort a production team will go through to avoid it. I’d be interested in knowing what tips the balances one way or another. When it is more cost effective to build special sets, rigs, etc just to avoid using special effects, and when chroma-key and CGI become more effective than trying to build it in real life.

Has anyone else wondered if Bruce Schneier’s “Friday Squid Blogging” is some sort of Steganographic message delivery?

Me neither.

The New York Times is reporting that Kevin Eubanks is going to leave the Tonight Show soon after Jay Leno returns as the host. It really doesn’t matter that much to me, since I stopped watching the Tonight Show soon after Branford Marsalis left the bandleader position.

In the article Microsoft starts over in phone software the writer Ashlee Vance says “The product marks a rare moment when Microsoft scrapped previous versions of its software in favor of building something new from scratch.”

I don’t see that as a very rare thing. I see Microsoft as a company that is very willing to scrap existing products and technologies to start over. Some that I can think of are:

• Multiplan to Excel
• MSN the non-internet “online service” (compare to Compuserve GEnie, Prodigy, etc) to MSN the internet portal. (and with that Internet Studio/Blackbird and abandoning all the developers they signed on to develop for it.)
• Windows 1.0 to Windows 3, (and for that matter Windows 3 to Windows 95) were probably as big of a jump as Windows Mobile to Windows Phone 7.
• MS-BASIC to QBasic to Visual Basic to Basic.net
• COM to .Net.
• Project Longhorn to what eventually shipped as Window 7.
• Microsoft Play-for-sure to Microsoft Zune.

Or for that matter, the long path that the PocketPC (for the PDA market) has taken to to the Windows Phone 7. Any other examples that anyone can think of?

The sentences before and after this quote seem odd to me as well first he says “Microsoft is trying to draw attention away from the application model and focus more on software that’s closer to the company’s roots.” and then “Microsoft has spent the last 18 months trying to add gloss and sophistication to a product that had suffered ridicule as being clunky and too wedded to the company’s personal computer roots.” Which is it trying to move closer to the company’s roots? Or trying to add gloss and sophistication because the existing product is too close to the companies roots? It seems odd to me that something that is described as being the “application model” is straying away from the Microsoft company roots. (Early on, Windows vs. MacOS was described as MacOS being “document centric while Windows being Application Centric. OpenDoc was designed to be Component centric to be a move away from Windows application Centric approach.)

(update on Feb 16: my friend Chris pointed out that I used the wrong gender for the writer Ashlee Vance: s/she/he/g)

Sometimes words have an air of trendiness to them. When they do, people start using them in a context beyond their original meaning, just to catch on to that trendiness. At that point either the word loses its meaning or a new word will come up to replace it.

Todays example: “hyperlocal” This article in the Boston Globe: Spilling the beans on McDonald’s coffee campaign describes their ad campaign as hyperlocal. Most definitions of hyperlocal that I’ve heard describe sites like Everyblock.com or Boston.com’s latest revamp of the Your Town feature. Basically the kind of stuff that Adam Gaffin has been doing in UniversalHub before there was a term to describe it. The McDonald’s ad is aimed at all of New England. How can that be considered hyperlocal?

I’ve heard about Yahoo Pipes a while ago, but it just occurred to me today that it could help with a few tasks that I’ve been meaning to work on. (the gtkpod documentation suggests that if a podcast’s RSS feed doesn’t work, to make an identity function pipe to fix it. I guess Yahoo’s RSS reading code is more robust than gtkpod)

My first task: taking the boston.com BostonUpdate twitter feed and make it easy to follow in an RSS reader. Boston.com has many syndication options, but they often have some flaws to them. Many of the RSS feeds are automatically generated, so they can be a bit noisy and low in content to noise ratio. Also, much of the content partially overlaps, so finding the optimum collection of feeds is tough (not to many duplicates across feeds, not too many useless articles, but enough to find all of the content I want to read)

The Twitter feed takes little advantage of what Twitter provides (very few hashtags. very little linking to other users.) but the RSS feed that the twitter account does have the advantage of being generated by news producers. The only drawbacks to the Twitter RSS feed are that posts tend to use URL shortening services like bit.ly, and that the links go back to Twitter, rather than the original article.

Yahoo pipes winds up being a convenient way of interacting with Twitter to get an improved version of the BostonUpdate feed. The Twitter RSS Feed pipe will take a Twitter ID, fetch the RSS feed for the page, convert all bit.ly links to the URL they point to, make the RSS link into that page (rather than the Twitter Tweet page) and turn the result into a new RSS feed. All in under a dozen statements (which Yahoo Pipes displays as flow chart style blocks that connect together.)

The bit.ly URL expansion was done as another Yahoo pipe (so I could publish it independently of the Twitter stuff and reuse it.)

My next tasks will be something that can take multiple RSS feeds and dedup them. That way when the same article shows up in the Boston.com most popular, Boston.com top stories, and the BostonUpdate feeds, I only see one of them.

I just read Top 10 Lies Newspaper Execs are Telling Themselves and it seems to start off as “What is the opposite of what the Boston Globe is doing”. The tips are: “We can manage this disruption from within an integrated organization” where more of Boston.com reports to Boston Globe management than three years ago. “Print advertising reps can sell online ads too” and the Boston.com sales teams were merged with the print sales teams about four years ago. “We can re-create scarcity by putting up pay walls”, we’ll see how that goes.

Judy Sims is an online media professional, so she may have a particular point of view. I guess over time we’ll see which direction works.

Boston.com announced a redesign of their homepage. They describe it as a “cleaner, leaner homepage” Since I had copies of their previous homepages for my previous experiment (fetched from the Wayback Machine), I figured I’d take a look at what their homepage size has looked like over time. The size of just the index.html file (not counting embeded images, ads, etc.) looks like this:

and the homepage as I write this is 103,478 bytes. The big jump in 2008 I’m not sure if it was short term thing. (there were only four days of data for 2008, so it may not be representitive) and I don’t have any recent data, but the most you can say is they brought the homepage back to 2001-2006 levels.

I’m sure that keeping the homepage under control is a daunting task, politically. According to the boston.com mediakit the homepage gets about a third of the site’s total pageviews. I’m sure there is a strong motivation for all the editorial departments, and the advertising department to add just one more thing onto the page.

It seems in the past month or so, newspapers have been making some noise that they are going to start charging for at least some of the content. One of the reports I heard about for boston.com was that people were considering charging for the content found in the Boston Globe, and leaving the rest as available to all readers.

That got me thinking about how much of what gets read from boston.com is Globe content, and also about how that may have changed over the years. I started thinking how traffic gets to the article pages anyway, and to a great extent the biggest traffic driver to an article page seems to be the homepage. This could be something I could measure, how much of the boston.com homepage is allocated to links to Boston Globe articles, and how much of it is from other sources. And has it changed over time. (My assumption was that Globe content would increase around 2006 when the boston.com and Boston Globe editorial departments became more closely integrated, and Globe reports started writing more direct to online content. Before that, the sense that I got was that Globe content was frequently highlighted for its unique view, but when news of the day changed from what was published the night before, the site started putting more up to the minute wire content from AP or Reuters.)

So I grabbed every copy of the boston.com homepage that existed on the Internet Archive Wayback Machine. I wrote a small script that would read each file and judge each link to be either a Globe link, a non-globe link, or one that didn’t count (I omitted things links to other section fronts like the news, sports, etc. pages.) I called a link a Globe link if it either was within a section of Globe content (/dailyglobe2/world/…) or the tease associated with the link had an attribution of “(Boston Globe)” or “(Today’s Globe)” I ran my script over all the homepages, put the results into a spreasheet to graph them and found….

Practically no difference in the ratio of Globe content vs. other content from about 2001 through 2008 (the last dates that the wayback machine has data.) Now I’m wondering what to do next. There may be something that I’m missing (I’m counting any link with a greater relevancy over another, even though links “above the fold” tend to get clicked on more than links near the bottom.) It could be the script I wrote to parse the homepage into globe/noglobe links isn’t counting things accurately. Maybe the data is right, but I should start playing around with it in r just to learn it.

Oh well, I was looking forward to publish the results that supported my hypothesis. Saying that I can’t support it isn’t nearly as much fun, but I figure its as much of a story to tell as the other.

I just read about a site called The Book Seer which does book recommendations. So I went back through the last three books that I finished (which were all very different from each other, so I wanted to see how it would respond:

Fun Home oddly enough couldn’t get any book recommendations from Amazon. Some of the recommendations it did find seemed to be geared more on its graphic novel-ness and less on its content (can someone see a connection Frank Miller’s Sin City and  Alison Betchel’s Fun Home besides they both have lots of pictures?)

Angels & Demons suggested most of Dan Brown’s other works, which isn’t very surprising. (my brother lent it to me well before the movie came out, and it just reached the priority level in the pile. We were discussing Bruce Scheier’s post about Hacking a Papal Election and the connection between security and tradition. I also meant to read it before I saw the film, but now its too late, at least for the theaters.)

Sacred Attunement was the oddest of all though. I got a whole bunch of recommendations labeled “attribute error at line …”.