Etech Day 2: Data Visualization and Vegetables
by David Sklar
Keynote from Tim O'Reilly
Tim's keynote touched on the "Google/Amazon/etc. as open source app" theme that he discussed also at OSCON last year (and he mentioned its familiarity) but also toured other areas of notable/high-potential/otherwise worthwhile activity that (he thinks) hold great promise:
- Hacking (in the tinkering sense)
- Second generation network effects like social software and net-enabled market research and data visibility
- The architecture of participation (e.g. the user-supplied content in Amazon search results)
- Getting beyond the single device
- Robotics and hardware hacking
The talk was a great "heads-up" for areas to watch. Some of this stuff I was familiar with, some I was just aware of, some was completely new. Smoothing out that "lumpiness" of knowledge is one of O'Reilly's stated corporate missions. Tim's talk was right on target.
Data visualization is an undercurrent to many of the themes touched on. For example, the sector maps available at http://smartmoney.com/sectormaps and http://netscan.research.smartmoney.com/Static/treemap. I hadn't seen these before and can think of a lot of other kinds of data I'd like to organize with this kind of tool.
Evolving the Bad Guy
Eric Bonabeau gave an entertaining and informative introduction to evolutionary programming. If it is axiomatic, as he claims, that:
- Our brains are bad at dealing with complexity.
- It is inevitable that a complex system has loopholes which can and will be exploited.
- Our world is filled with complex systems (tax code, power grids, software, etc.)
Then, harnessing the power of evolution in programming gives us a way to discover loopholes in these complex systems that we couldn't find otherwise. The basic idea is to create a symbolic representations of actors in a system, a fitness function that measures how that actor performs, and the rules in a system. Then, recombine the pieces of those actors and evaluate their fitness to evolve the most fit actors.
The key benefit of this technique is that it can evaluate a huge number of potential solutions -- far more than any more traditional method could. It also evaluates solutions that aren't thought of a priori by developers or testers. It's sort of like running all computationally possible unit tests on a piece of code without explicitly writing any of the unit tests.
Next Generation File Sharing With Social Software
Robert Kaye laid out how to build (with plenty of prior-art worthy detail, patent hounds) a file sharing infrastructure that provides sufficiently adequate protection (in the worst case compromise scenario) against identifying network users with anything else than anonymous IP addresses.
The technologies he talked about are not revolutionary (which he acknowledges) -- a combination of web services, SSH, PGP, and existing P2P protocols and clients. The distinctive part are the social models. His envisioned infrastructure lets different "tribes" set their own rules on what it takes to be part of the group and what kinds of privileges you have. Individual tribes can similarly define how they relate to other tribes and so-on, forming larger aggregates of interrelated individuals which can still preserve their local rules on interaction.
The weak links in a crypto or security regime, Robert emphasized, are people. For example, someone who shouldn't gets permission to join a group. So the solutions to these kinds of problems involve the rules for interpersonal interaction and having "tribal leaders" establish and enforce those norms.
I don't have a mobile phone, so I went to this talk to see what I
am or am not missing out on. I'm not now going to run out and buy one,
but what this session really hit home for me was that voice
communication was a popular, useful trojan horse which got millions of
people to start carrying around pocket-sized (or thimble-sized, if
wirelessly networked computers.
A lot of what's being presented all over this conference are
brainstorms of what to do with this swarm of chatty, roving
computers. These brainstorms are all over the
creepy/exhilarating/neat/silly map. I am really curious not only to
see how the technological problems and solutions evolve but also how
the cultural/etiquette ones do. Perhaps someday soon my (hypothetical)
phone can glow red when I'm walking by a bar in which there are people
who I'll probably think are cool and blue when I'm walking by a bar
that I should avoid. Now imagine I'm outside a bar with five other
people who have similar phones. Three of our phones are red, three are
blue. Is it polite to go in? Should I be embarrassed that my phone is
blue and a friend's is red (or vice versa)?
Etiquette for talking on a mobile phone is jumbled enough as it is:
everyone seems to think that talking on the cellphone around others is
rude but they still do it. It's a nono for a phone to ring in a
theater but it happens all the time. (And it's illegal to have a
device in the theater that jams mobile phone signals.)
So, all of these neat apps raise all sorts of other neat cultural
issues that I look forward to seeing be hashed out.
Folks from Ludicorp talked
about snazzy stuff they've done to enable relationships formed in
online MMOGs to extend into the broader net. For example, some
send messages to you inside the game (if you're playing at the time).
The Q and A part of the session provided a stellar reminder about
what's great about conferences like this: someone asked a question
about FOAF and the answer involved a discussion of permissions issues
related to importing a list of friends into a social networking web
site. When answering the question, Stewart Butterfield, as a random
example, said something like "You might import your friends list that
says you're friends with Jeff Bezos, but Jeff Bezos would say 'Who the
heck is this? I don't want to be your friend!'". At that point, Jeff
Bezos, who happened to be sitting in the front row, got up and said
something like "I'm sure that would never happen." (to great
laughter). It was sort of like being an extra in the movie theater lobby scene in Annie Hall.
I cut out at lunchtime to drive up to the
Farm and get some vegetables. I got some great strawberries, small
white turnips, and a few kinds of radishes. If you want to go, get on
I5 North from downtown and go about 19 miles to the "Via de la Valle"
exit. Turn right at the exit on to Via de la Valle, go about 4 miles,
and turn right onto Calzada del Bosque. The vegetable shop is on the
right about a half mile down the road. It's about a 20 minute
drive. Before you go, call 858.756.3184 and listen to the recorded
message to check their hours and make sure they're open.
One of the best things about a conference like this is the
opportunity to meet and talk to so many bright folks doing neat
things. Most everyone is pretty friendly, too. When someone's given a
talk or been on a panel and I later want to talk to them, here's my (so far
effective) strategy: find them wandering around in the hallway (or stand nicely next to them if they're talking to someone else and it seems like a general chatting-at-the-conference talk, not a personal conversation), tell them something I liked about their talk or
what they said on the panel. Then ask them whatever question I wanted
to ask them. This usually leads to an interesting and/or educational
conversation for a few minutes. There's no technology that mitigates
this -- it just comes down to good ol' introducing yourself to
Although many of the sessions are full enough that this effect is camouflaged, attendees tend to sit where there are power outlets: along the walls and near clusters of powerstrips placed here and there amidst the seats. Tim's keynote mentioned The Word Spy, which elucidates the meaning and first use of new words and phrases. So what's a new word/phrase that describes this seating preference in which conference attendees arrange themselves like iron filings on a piece of cardboard with a few magnets underneath it? "power seating"? "strip seating"?
What's it called when where you sit is determined by where you can plug in your laptop?
Sklar writes that ´there´s no technology that mitigates his strategy to meet people: ´find them wandering around in the hallway or stand niceley next to them, and tell them something, he recommends us. Sklar probably has never heard of the swarming-technology of eyebees.com: software that enables people to get aware of each others presence, meet and talk to them; real time and while on the same location, web- or filebased.
You're absolutely right -- I've never heard of this swarming stuff, which seems to require a PC running IE 5.0 or greater, new software to be installed, and me to be typing away at my computer to interact with people.
Open Source Tree Map Visualizations
If you want to play around with tree maps like those used by Smart Money and Netscan, check out the open source tree map my company is releasing at: