Anirvan Chatterjee

I had a full-on hard drive crash a couple of weeks back. One day my Mac Mini worked. The next day it didn’t. I banged on the computer every way I knew how, but to no avail. I finally took it down to my local computer store, where I was told that there was a hard drive crash. It was OK, I told myself. I have full backups.

About two years ago, B and I decided to work on a backup plan. We have too much important data on our computers (photos, documents, etc.) to see it go away one day. I bought a copy of Take Control of Mac OS X Backups, and waded through the options. I bought an external hard drive, hooked it up to my Mac, and tried to turn it into something like Apple’s Time Capsule (which didn’t exist at the time). That didn’t work out so well, but Mozy did.

Mozy is an online backup service. The software backs up your data in the background, over the net. Because it’s online, it’s as slow as your net connection. In my case, it took about three days of continuous online data transfer to get everything transferred; afterwards, the software runs in the background, updating modified files. It works great, and costs just $5/month for unlimited backup (in my case, almost a hundred gigs of data).

B’s laptop suffered a hard drive crash almost as soon as she started using Mozy two years back; she had most of her her data backed up, and we got most of it all back. I had everything saved, and got everything back. Mozy’s restore process is a pain; while you can restore individual files immediately, it takes days to prepare dozens of gigs of data for download. Then you download the data, which, for large quantities of data, can also take days. And the restore process is a bit clunky, to the extent that I had to write a Perl script to make it work the way I wanted it to. The process also took longer than I’d liked, so I’m probably going to add SuperDuper to my backup mix for the future.

But it worked—I had a huge crash, and within a week, I had all my data back the way it was before. That’s pretty amazing.

I’ve been enjoying listening to Spark, a Canadian (CBC) radio show on technology and culture. I’ve been listening to their summer long-form interviews, getting a chance to hear from folks whose work I know mostly from blogs and print:

• Rick Prelinger, Prelinger Archives (MP3)
• Alex Steffen, Worldchanging (MP3)
• Xiaochang Li on Soulja Boy (MP3)
• danah boyd and Marc Fisher (MP3)

I just discovered CBC Radio’s podcasts. Besides Spark, I’ve also been listening to:

• C’est la View, an English-language program on Francophone culture
• White Coat, Black Art, a program on medicine and medical practices and culture
• Canada Reads, an annual reality-show-style competition where celebrities duke it out to promote their favorite Canadian book

When I first started working, I decided to give away about a month’s income (about 8% before taxes) every year to nonprofits. Working in the tech sector during the dot.com boom, I figured I’d always have more money than time, and that I could be most useful bankrolling groups doing work I believed in.

It was relatively easy at first. My costs were low, certainly low enough to give away a bit. I think back to how freely I pulled out my checkbook at the time, and how much I gave without thinking about it. As I got older, costs started going up, and I realized that I needed to get serious saving for the future. At the same time, I realized I enjoyed volunteering, and that small groups sometimes need manpower more than they need money. The more time I gave, the more my donation rates slipped. B and I got married, we bought a house, and took on a mortgage.

As we were doing our 2006 taxes, we realized that as two white-collar working adults without dependents, we’d only given away about 4% of our income in the past year. It was embarrassingly low. (The average household earning under $10,000 a year gives away 5.2%.)

Over the past year, we’ve been working to change that pattern, giving away more to groups we already donated to (e.g. Asha for Education), and becoming first-time donors to groups we admire but have never supported financially (e.g. Stop Prisoner Rape). As I finished our 2007 taxes in April, I was happy to see that we made it to 10% in the past year.

I’m inspired by the stories at Bolder Giving in Extraordinary Times, which advocates for substantially higher levels of giving by those of us privileged enough to do so. (Which, per the Global Rich List, is pretty much everyone I know.) I’m not there yet, but it’s something to aspire to.

I married into Netflix. I never used to rent movies, but B had an account, but when she gave me a profile under her 3-DVD plan, I started discovering a world of great movies that I’d missed in the theater. The more movies I rated, the better Netflix’s collaborative filtering performed. By the time I’d rated over 600 movies (basically every movie I’ve ever watched), I found myself coming to really trust the Netflix engine, to the point that I now rarely ever watch a movie unless Netflix projects that I’d rate it a 4 or higher. The predictions are eerily accurate—and help me avoid wasting hours watching bad movies.

I was disappointed when Netflix announced that it would cancel the profiles feature, allowing a single account to be split up into multiple queues . Without profiles, B and I would have to end up sharing queues and ratings, giving both of us crappy recommendations. I was particularly annoyed that Netflix would delete all my data.

Annoyed with Netflix I developed a tool last month to export user profile data, back it up, and optionally upload it to another Netflix account. Thankfully, Netflix relented soon after, so the immediate data deletion threat was over. I’ve been using the opportunity to experiment with moving my ratings data to other recommendation platforms, to combat lock-in and test other engines.

The first alternative movie recommendations platform I tested was Yahoo! Movies, powered by ChoiceStream. Porting hundreds of ratings over from Netflix to Yahoo was more difficult than i expected. Yahoo didn’t have every movie that I’d rated in Netflix; in many cases, it was difficult to disambiguate between different works with similar names.

After transferring several hundred ratings, Yahoo! Movies spat out the following recommendations for movies for me to see at the theatre:

1. The Visitor (hadn’t heard of this, looks interesting)
2. The Fall (Tarsem visuals, but iffy plot?)
3. Brick Lane (saw it, glad I went)
4. The Dark Knight (planning to watch)
5. The Incredible Hulk (ick, no way)
6. Kit Kittredge: An American Girl (do they think I’m a 9 year old girl?)

I really have to wonder about The Incredible Hulk and Kit Kittredge: An American Girl, but the first four recommendations aren’t bad. Unfortunately, Yahoo recommendations only touch movies in the theater and DVD releases, but they do do a good job of providing an immediate personally-filtered read on what interesting general-release movies are playing. I plan to look them up the next time I want to go see a movie and have no idea what’s playing.

I’m reasonably happy with this project so far. I’ve been able to get my data out of a closed system, save it as a personal backup, and move it to another platform. If Netflix had kicked users like me off their site by shutting down profiles, I would have had the capacity to move. While Yahoo! Movies’ collaborative filtering system isn’t a reasonable substitute for Netflix’s at this time, there may be others out there that do a better job.

I was walking near home Sunday when I saw two 25-35 year old normal-looking white guys walking by, one of them with a tattoo in Indian script on his upper left arm walking by. As we approached, I glanced at his tattoo again. Jeepers. It wasn’t Devanagari, but Bengali script. Who walks around with a Bengali tattoo?

By then, he’d walked past, but I turned around, and called out to him, “hey, excuse me, can I read your tattoo?” He smiled, extended his arm, and I read “Hare Ram Hare Ram Hare Ram” in a circle around his arm, an invocation to Rama, the mythological Hindu warrior-king. We introduced ourselves, talked for a minute. The tattoo bearer described himself as a Vaishnavite (a strain of Hinduism popularized in the west by ISKCON), and was aware of the Bengali language program at UC Berkeley.

We talked for a minute more, shook hands, walked away. How random that the first person I’d see with a Bengali tattoo would be a non-cultish-looking white Hindu guy in Berkeley.

I don’t believe that marriage is the single most critical public policy issue that LGBT Americans need to be fighting for right now; I suspect, for example, that expanding health care access and the reach of anti-discrimination law may have broader positive impacts. I’m also not sure that same-sex marriages will have better outcomes than opposite-sex marriages (about half end in divorce; many harbor emotional or physical abuse).

That said, the visibility and growing success of the marriage equality movement is remarkable, in the way that it helps reframe homophobic policy discourse into a parody of itself:

• 1970s: gay predation endanger children
• 1980s: gay diseases endanger public health
• 1990s: gay demands endanger military cohesion
• 2000s: gay marriages endange straight marriages

Marriage equality campaigners do a good job of heightening the contradictions, forcing homophobes to make increasingly convoluted arguments as to why it’s bad for people to have stable, monogamous, relationships saddled with legal rights and responsibilities. Since LGBT folks can’t be wished/prayed out of existence, the movement forces opponents to essentially say “down with gay marriage; I’d rather have more frightening, irresponsible, uncommitted, unmarried gay sex.” There’s something delicious about that.

I was at the BART station from 7:00am to 10:00am this morning, handing out last-minute voter pamphlets for Kriss Worthington, a local Democrat running for California State Assembly. The primary election’s today, and in a heavily Democratic district, the winner of today’s primary election is almost guaranteed the seat in November.

Leafleting is fun. I did a fair amount of it in college, around Prop. 209 and other issues. I like how it gives you an opportunity to connect with people in public places, and sometimes start conversations you wouldn’t normally otherwise have.

I’ve found that getting someone to take your leaflet is a little three-step act of marketing, performed in anywhere between one to fifteen seconds:

1. Don’t automatically turn people off
   People make snap judgments about whether they want to walk toward you, or away from you. I try my best to look approachable and respectable when leafleting. It’s hard to make a pitch if the recipient’s already biased against you.
2. Make the pitch
   When leafleting, I call out something intended to catch the interest of a passerby, so I can offer the flyer. I usually try out a few pitches, and see which works best. This morning, I started off with “Kriss for Assembly,” but his low name recognition in a little-covered race meant not many people connected. After a few more attempts, I found the most success with “don’t forget to vote,” and sometimes “Democrat or independent?” When I ran out of Kriss Worthington materials, I moved on to handing out flyers against California Proposition 98). Selling that was much easier, since it’s a well-publicized statewide issue. I was most successful with “remember to vote / no on 98” (occasionally customizing the pitch to insert keywords I thought might be of interest, e.g. mentioning tenant rights to college students).
3. Help the customer understand what you’re offering
   Once I’ve gotten someone’s attention, I have maybe a second or two to convince him or her to take the piece of paper I’m handing out. I try to hold the leaflet out at mid-chest height, angled so that readers can easily see what they’re being offered, and placed within a foot or two of their walking path so they can easily grab the copy being offered. This is tricky, but incredibly important. Unless they happened to connect very strongly with your pitch, most passersby won’t make very much of an effort to take your flyer; they need to be quickly reassured that it’s safe, useful, and easy to take.

Unsurprisingly, the marketing of leaflets has a lot of parallels with marketing our web business:

1. Don’t automatically turn people off
   Running a web business, that means having a reasonable name and URL, and a fast-loading and non-overwhelming website.
2. Make the pitch
   Since we buy short (95-character ) online text ads to publicize our site, we spend a lot of time thinking about how to deliver our message in a small space. We’ve tried out hundreds of variations of our ads against an umpteen number of real-life user searches, working with the ad tracking software to determine which pitches work best for each search keyword and audience segment.
3. Help the customer understand what you’re offering
   As a web business, this means using design, branding, and instructional and trust copy to help convince users that the site’s a good fit—safe, useful, and easy to use.

Looking forward to seeing the California state primary election results come in tonight…

Update: The results are in:

• WIN: California proposition 98, sneaky eminent domain reform which would kill tenant protection, lost (39%-61%)
• WIN: California proposition 99, a clean eminent domain reform bill, won (62%-37%)
• LOSS: Kriss Worthington, my preferred Democratic candidate for State Assembly, lost to Nancy Skinner (16%-47%, in a four-way race)

I was one of the judges for UC Berkeley School of Information’s 2008 Master’s final projects competition this afternoon, helping pick the best among the projects in the information systems implementation track. We saw several really interesting student projects. Most of the presenters held up under the heat and stress, and I enjoyed seeing the level of attention to detail that the best project teams brought to the table. Thankfully, the judging wasn’t all that difficult; Peter Merholz, Jeff Ubois, and I came to a decision pretty quickly. The winning project team will be informed during their commencement ceremony tomorrow.

Seeing the students getting ready to graduate was bittersweet. I took a leave of absence from the school’s masters program nine years ago. Getting into the program, then called the School of Information Management and Systems (SIMS), was all I wanted to do with my life. I was admitted as part of the second entering class, but eventually left school to try to launch a startup around a class project that was gaining traction online. Though it was clearly the right choice to make, I still feel like I gave up on something really important. I’m glad to be able to be a part of the school’s extended family, through links with past professors, fellow alumni, and current staff.

On Saturday, I went precinct-walking for Kriss Worthington, a Berkeley city councilperson now running for the California State Assembly. I went door to door around town, talking to neighbors about Kriss and his campaign.

Here’s what I learned:

• Very few people want to talk about political candidacies on a Saturday afternoon.
• Of those that do, most don’t follow the state assembly race.
• Local politics matter; one man told me he’d never vote for Kriss, because of his position on a local park issue.
• Voters are interested in more than data. I found myself deviating from the suggested talk points on issues and endorsements, and talking about why Kriss made an impact on me and why I was out walking for him. I felt like I swayed several likely voters during a couple of long conversations.
• It felt really good to know I wasn’t alone, as I met several other intended Kriss voters.

It was my first time precinct-walking, and I was paired up with a sixtysomething white woman. As the day went on, we realized we shared something more than our support for the same candidate — a common address. It turned out that she used to live in our house, renting the condo right underneath ours for several years through the 1960s; she cared for several of the trees still in our backyard. We enjoyed catching each other up on the building and the neighborhood, and made plans to meet up again. Small world.

I’ve made a few tiny contributions to open source tools over the years, but nothing significant. I recently uploaded Business::LCCN, my first public Perl module, and open source contribution under my own name.

Library of Congress Control Numbers, or LCCNs, are codes that point to book records in the US Library of Congress. I’ve been encountering LCCNs while working on a personal book cataloging project. There are some weird ins and outs to dealing with older LCCNs; I had to contact the Library of Congress to learn some of the obscure details that aren’t publicly documented. Business::LCCN parses and manipulate LCCNs, taking these special rules into account, abstracting away most of the complexities.

Perl modules uploaded to CPAN get run through a variety of systems:

• public code repository
• bug tracking
• automated test reports across a variety of platforms
• dependency checking
• reviews
• code quality metric checking

I’ve really enjoyed being able to write some code, and get so many other services for free.