(this is page 97 of 121)





Latest

My friend Jerry Michalski has long held that businesses that think about "consumers" are setting themselves up for failure from the start. In what will stand the test of time as a brilliant quote, he stated that companies who thought about the "consumer" reduced him to "a gullet whose only purpose in life is to gulp products and crap cash." Now Josh Bernoff from Forrester is arguing that the term "user" has been equally debased and that we should all just stop using it.

The term "user" in the "consumer Internet" space has taken on special importance as a measure of value. MySpace, Facebook, YouTube, etc. are all measured in terms of "unique users," and every little company looking to make a similar run has a board full of investors asking the same question every month: "how many users have you got?" We certainly did.

The bummer about it is not the use of the term— but rather how much the constant focus on growing the pool of users quickly distorts any connection that the word user actually has to a real breathing person, even in startups that claim to be "user-centric." You stop focusing on the individual people coming through the site and start abstracting away to the clickstream, the userflow, or the suite of A-B tests. All of these tools have their place to be sure— but they tend to further exacerbate the problem brought on by the constant question at all of the board meetings: "how many users?"

In his post, Bernoff cites an essay by Don Norman (a personal hero) who argues that it is really important to focus on the language used to refer to the users of a product or service during the design process. The piece is required reading for anyone who works on anything meant to be used by anyone else, as it will begin to make you aware of all of the in-built biases that the all of the associated practices around user experience have.

If you like the piece, I'd go out and get his book, Emotional Design, which rocked my world when I read it with the seemingly trite starting premise that each of us in just a big bag of emotions and that the sooner we design products that accept and leverage this, the better off everyone will be. For the past year now I've been grinding through cellphones wondering how it is possible that the folks at Motorola, Palm, Nokia, and now Apple, haven't all inhaled and internalized the messages in this book. After all, what is more emotional than the way in which we communicate with the folks we care about the most?

I'm all for dumping user but I'm sure that it won't be long before some other term comes to replace it— author, owner, creator, or perhaps some Sterling-esque made up word. To get rid of the problem of abstracting away the person behind the wallet, the human behind the eyeballs, it takes making their own experience using your product or service the end goal and not just the means to something other goal. And recognizing us as the emotion bags that we are might be a good place to start.

Comments Permalink

Building Businesses in Boston

I first met Scott Kirsner back in 2001 when he wrote a column about Memora, a company my brother and I had started at the intersection of consumer technology and broadband infrastructure. I liked Scott because of how much he seemed to "get" what we were trying to do, and how far ahead of other journalists he was willing to see. Naturally I was disappointed when he moved west and subsequently glad to see that he's come back, though this morning's Sunday Boston Globe article on how we're just continuing to miss the consumer tech opportunities here on the east coast rankled me a bit. It reminded me of Mike Arrington referring to Boston as a backwater, and even more so, of all of the folks I would meet on the west coast while traveling for Tabblo that argued for moving the entire company out there.

For the record, right before starting Tabblo I almost pulled up stakes and headed to the west coast for most of the reasons given in Kirsner's piece. That I stayed is a tribute to two things: my faith in the depth of the Boston-based engineering talent that I'd been building into my network, and the good efforts of the folks at Matrix who single-handedly funded the plan with little more than a concept, me, and a sales pitch. Which is not to say that it was not a painful process to make the financing happen: it took a lot of education and hard thinking on both of our parts to come to the conclusion that there was a viable, scalable, and ultimately defensible business model in what Tabblo was going to do. But when after just under two years, HP came knocking, I was very glad that we had done the hard work up front and that we had the depth and breadth we needed on the team.

Scott is certainly right about one thing— starting funded consumer Internet "Hail Mary" businesses in Boston is pretty nearly impossible. There is no VC in Boston that I know who would have funded YouTube, and for good reasons too; namely, cost structure and copyright. But those Hail Mary plans rarely succeed and for every YouTube, we've got several Napsters to prove the counterpoint. Instead, we've got the less sexy but more fundamental consumer Internet businesses here: Zipcar, Kayak (ok, in CT), and TripAdvisor to name a few. I know folks at all of these companies and they all share an incredibly analytical view towards what is happening on the "Internet as distribution channel" which is simply not a part of what I know of the west coast ethos.

I certainly do not mean to knock the swing-for-the-fences mentality of the west coast, and as an entrepreneur, I would certainly kill for the recruiting efficiency of being able to drive down to Google or Apple for some talent. To boot, we've still got a lot of Digital/Lotus detritus to clear out of of the VC pipelines before we can really adequately fund the new opportunities the Internet presents with the proper teams of folks. But give us time Scott— don't lose faith quite yet!

Comments Permalink

Lots of people have been spending way too much time blogging about the iPhone (though believe it or not, I still think it is under-hyped), so I'm going to stay away from reviewing mine. That said, I did want to make a quick software comment that the iPhone's keyboard makes clear as day.


Whenever you can turn a piece of complex UI from hardware to software, you seriously accelerate the rate of improvement of the general interface.

I've had two other Smartphones, the Treo 700p and the Nokia e61i. Both have the full tactile keyboard that the RIM Blackberry pioneered— which despite looking like it was made for Smurfs, works remarkably well. However, you make two key compromises with it: the loss of real estate that could either go to the screen or to a smaller overall device, and a hard upper limit as to how fast you will be able to go. In the case of the iPhone, it seemed to me that Apple's engineers valued screen over buttons and went with the soft keyboard as a compromise.

And despite the overwhelmingly positive reviews of the soft keyboard, I never in a million years expected that I would actually find myself typing faster on the virtual iPhone keyboard than on a tactile one. Yet three weeks into it, that is exactly what has happened— I can now type faster on the iPhone than I ever did on the Treo or the Nokia, thanks exclusively to the interface and to the simplistic spelling correction algorithm. More importantly, the rev cycle just went from 12-18 months (the average amount of time between my hardware updates) to whatever the frequency of the iPhone software updates shakes out to be. They've taken something that was tied to the limitations of the physical world and moved it squarely into the realm of the virtual.

ReadWrite web had a thought provoking post on exactly this topic, "digital physics," arguing that an interface which borrows just enough from real world physics but takes it beyond those constraints is the future of all user experiences. At the time I saw it as an esoteric point, but now that I've experienced it, I'm completely sold.

And of course there are other benefits to moving the interface into software. When Tivos replaced VCRs we got a big boost in "recording productivity" just because of how much more heavy lifting was being done in code. I suspect we'll see a lot of the same types of things in the iPhone's virtual keyboard, especially as the API is opened up to third-party developers.

That said, there are clearly examples of when moving an interface from hardware (buttons and knobs) to software can really screw things up: BMW's iDrive and all of those fancy universal remotes being good case studies. It would seem to me though that most of these counterpoints are better examples of poor implementations and not so much a weakness in the idea of moving hardware interfaces to software.

One final caveat: there is this famous Alan Kay quote about how people who are serious about software ultimately build their own hardware (which incidentally Jobs keeps using about all things Apple). I've never understood it until now. It is after all the incredible sensitivity of the panel in the iPhone that lets the software do its magic, and we'd do well to remember that bit of wisdom before even considering the approach.

Comments Permalink

I have been a fan of Jon Udell's ever since his awesome "Practical Internet Groupware" hit the stands back in 1999. So when he interviewed the dean of the CS department At Carnegie Mellon, Jeanette Wing, about her views on the importance of "computational thinking" as a core discipline that should be taught to kids early on, I was intrigued. Lego Mindstorms and Milton-Bradley BigTracs notwithstanding, I've always found the ability to reason abstractly and algorithmically about a problem a huge asset that has been helpful to me way beyond of the realm of programming a computer. And now that I've got Alex (my five year old) getting to an age where he can start to grok this, I'm obsessed with taking every opportunity to pass this down.

Scrambled Squares

Which, as you can imagine, doesn't make me (or the rest of my family) that much fun for playing games or doing puzzles. Ever since I can remember, every game we've ever tried to play quickly turns into an excuse for thinking about how a computer might do it better: from Blackjack to Clue, from Stratego to Scrabble. This weekend we had a similar situation with a puzzle called "Scrambled Squares" from a company called B-Dazzle that my in-laws had bought for Alex because the puzzle was about pirates (a long standing obsession of his). On the face of it, the puzzle appears to be about spatial reasoning: nine squares, each with 1/2 of a pattern on each side, have to be put together to form a complete 3x3 super-square where all of the patterns "foot." But with way too many combinations, spatial reasoning and brute force quickly give way to tearing your hair out and declaring the puzzle unsolvable.

And thus, Scrambled Squares is a great puzzle for teaching the principles of computational thinking. The puzzle is mostly governed by very visible constraints which quickly force the issue of more abstract thinking, and to boot, it's got a great graphical interface that encourages all sorts of direct manipulation.

In the end it was unrealistic of me to expect that I could teach him how to solve it for himself (I could barely solve it myself!). But I was very pleased to see that certain lessons (find the middle piece and work outwards, find blocks of two and try those as 'units,' turn the highest 'constraint-pleasing' edges inward) were quickly internalized and then repeated to the gathering crown of befuddled adults. What we ended up with was much less of a formal algorithm (though a subsequent Google search turned up a whole term project on it), and more of set of rules to cut down the space of possible arrangements. But it is precisely this type of early computational thinking that I'm guessing does not get covered in primary school math classes and thus why we may indeed need to introduce a new top-level discipline right from the get-go.

Comments Permalink

Sorry to everyone that comes here for the "big thoughts" about the following shop talk. Feel free to skip— for those interested in web frameworks, read on...

Being the conservative guy that I am, I was willing to bet millions of dollars of investors' money on an "unproven" open-source web framework, but not my own puny personal blog. But a few days of vacation have finally cleared the head and have caused me to abandon the whole Ruby-on-Rails powered fad (more specifically the twice-abandoned Typo project) for a blog powered by Django.

I can't say enough good things about Django. Professionally, it was one of the best technical decisions that I got to make early on at Tabblo. I had always assumed that we would use it for the alpha, maybe the beta, and then dump it for a custom system that could perform. Instead we rode it from the first alpha until the day we got acquired by HP. In the meanwhile we ran 5 million photos through it, a million users, and over 150,000 content publishers. We had some ugly moments (the shift from mod_python to lighttpd/FastCgi and the move to magic removal stick out in my mind), but overall it performed incredibly well. A lot of props go to Adrian and gang for putting together such a flexible framework. They used it as a lightweight CMS, but we were able to take it and build out a full UG content publishing platform (with a lot of help along the way from memcached and solid engineering on the part of the Tabblo folk).

In fact, the HP folks doing diligence on us started by asking why we had not chosen Ruby on Rails and came out the other end incredibly impressed with our Django choice. We survived our security audit (coming out much more unscathed than a lot of other HP acquisitions) thanks mainly to Django, and have continued to develop on top of it even as we get ready to truly become the "print engine for the web."

For sure there are things, I'd love to see improved, namely: mutliple DB support (sharding), a better testing framework, and general speed improvements. But hey, now that we're HP maybe we can find a way to help out with those things!

A Django site.

In the meantime, theonda.org (or an.ton.io) is now officially "Django Powered." I haven't had this much fun programming since at least 6 months before we got acquired. In fact, I've decided that Django— when it comes to web stuff— is perfect for "vacation programming." What do I mean by this? It's the type of programming where you get to focus on the fun stuff and ignore all the crap you find yourself having to do again and again on every web project. A few examples:

* Generic views: if I had 5 dollars for every time I've parsed URL fragments to present a list of objects by date/type/whatever, I'd be long retired.

* RSS/ATOM framework: this one is a bit messy (tightly coupled) but man is it great not to have to deal with XML encoding/decoding. Wowee, the funnest 30 minutes of the project.

* All sorts of nice html-ifying functions: want to wrap your paragraphs in p tags? Format columns of text? All built in with Django.

Finally, Django is just well architected. I use a venerable Mac program called MarsEdit to write all of my blog posts. Unfortunately MarsEdit seems to support mostly crappy old XMLR-RPC-based protocols for publishing (Blogger1, metaweblog). Fortunately for me, between Python's support for xml-rpc and Django's URL dispatch mechanism, supporting metaweblog was a "nap time" exercise (it would have been less time if there was clean and current metaweblog docs on the web). Fun stuff.

Now that I've moved to Django, I'm really excited to start playing again. The thing about Typo was that it was hard to make any changes to it without having some of that RoR "magic" explode in my face. With the Django there is no magic— just clean, explicit, and simple design.

Well done guys!

Comments Permalink