29 March, 2007
Today just a few pictures from the Make: Fest at O’Reilly’s Emerging Technology conference 2007 (etech2007). The ‘fest’ was not that big, but there were some cool projects. Especially the 3D projects were awesome, all hacked together with fairly little resources and normal parts. Great stuff.
Here are also some pictures from the hotels I’ve been at this week. The Hyatt for the conference and the Marriott as my home base.
28 March, 2007
OK, the title of this talk was a bit on the vague side, but it did turn out to be very inspiring. The bottom line was how to keep complex systems usable.
Charles Armstrong took stage for the first part. His point was about making ‘sociomimetic’ systems easier to use. Sociomimetic stands for mirroring social behavioural patterns in electronic information systems. What it boils down to: the underlying system becomes complex and not intuitively understandable, but there are still people wanting to use it. How do you give the users some guidance for their intuition about how the system works.
Basically there are three factors to do this:
- grokability – make it easy for the users to understand what something is. Even a ‘cavemen’ should be able to understand what an axe could do.
- predictability – once the tool is understood, is it predictable in its function? If you understand what a computer can do, it is still not predictable how to accomplish certain tasks. This makes for a very steep learning curve.
- relevancy – having understood, being able to predict how a tool works, is it actually useful?
But how could these factors be addressed in complex systems?
- To get a better grokability something should be as simple as possible. Armstrong gave the example of two london subway maps. The newer maps (as we know them today) don’t really convey the real world, but they are really good to understand how the system is laid out, and make it possible to form a mental model.
- As an example of how to improve predictability the Eurofighter Typhoon Aircraft was given. Without any automatic adjustments this type of plane would be almost impossible to fly, because of its ‘aerodynamical unstability in the subsonic region‘. Below the standard aircraft control system is a very advanced system to keep the plane predictable. Again, this complexity is not exposed to the pilot resulting in a predictable experience.
- Although politics in a democracy is very complex, politicians seem to do a good job of conveying the relevancy of what they do. How they do it? By simplifying their messages to the bare minimum. They persuade voters with ‘in-your-face-usefulness’ like better education and lower taxes.
Mike Stenhouse took over the talk and showed a lot of examples of systems that were inherently complex, but easy to use. He started with the example of the power of photoshop filters: very complex, not intuitive. However, in the 1990s there were KPT filters, a break from the norm, which made filters very intuitive. Other examples were: the hidden complexity of 3D modelling in Bryce, tag clouds in last.fm that include authority of the tagger (but nobody should notice this complexity), google search (inherently complex, but just one text box to query), flickr.com interestingness, etc, etc. Bottom line: it IS possible to address the above factors in complex systems design.
Some hints and tips were shared at the end: It’s good to use metaphores. In their product (from Trampoline Systems) they use the radar metaphor. With a slider the range of the radar can be changed, resulting in less or more email about certain topics. Also, an expert mode is not always a good idea, because it might intimidate the average user. A better way to go around this is to slowly offer more functionality to people who seem to be experts.
Of course, this presentation sparked some questions from the audience. Someone asked if it would not weaken the tool if it was too simple. Another audience member actually answered with the example of the ‘choke’ in old cars: nobody misses that.
28 March, 2007
In a way I’m a bit disappointed by the talk of Jeff Hawkins on Hierarchical Temporal Memory (HTM). But, I guess, it’s my own fault as well. Already I read his book (On Intelligence), read up on the NuPIC platform and played around with the demo application. So what can you expect from a 45 minutes talk on the subject.
One thing that came up in the talk were limitations of the software. At this moment two things are not yet tackled. First, precise timing cannot be addressed by the system. This would mean that (near) real-time control of a system will be a challenge. Another issue is modeling natural language with HTMs. At this moment there is not a clear idea yet how to go about this. In general, Numenta does not focus on specific applications. Their business model is based on earning money with a licensing model. For demonstration purposes they focus on vision system, which is also shown with the released demo application.
In the end I’m still curious to see what NuPIC can do, and like to see another application other that the (still impressive) demo.
27 March, 2007
The point Jon Udell makes is exactly the weakness that was exposed on Etech2007. Robert Cook (freebase) gave a presentation about what freebase could be. THE problem with the whole idea is that they do not know yet how they are going to tackle a user driv
Jeff Jonas was part of the O’Reilly Radar Executive Briefing on Etech2007. His ideas were thought provoking. Mainly about identity, privacy and databases and how to combine data and queries.
27 March, 2007
Another part of the ‘briefing’ today was about how web 2.0 and wallstreet are related. Seems like a far stretch, but actually wallstreet turns out to be fairly connected (yes, that is an understatement). Nowadays trading takes place within (a) millisecond(s), and no that is not overstating it.
Peter Bloom and Bill Janeway took a seat on the couch, and rapidly fired interesting pieces of information about wallstreet into the audience. The pace was high and the concepts abstract. The main issue seems to be that trading information has become invaluable. This in turn means that a lot is being done to make sure that information does not fall into the hands of people (or bots for that matter) that shouldn’t know. The study in this area is called ‘behavioral finance‘.
Not just pace of this talk was high, but the pace of trading is rapidly changing. A couple of years ago guarantees for trades were in the order of seconds, now their almost at a sub-millisecond level. There are trade brokers, who operate at about 1000 trades per second. What happens is that the time to change a mistake (which sometimes happens) goes to zero. This has great implications in the world, where computers are taking over from humans.
Tim O’Reilly tried tying this back to web 2.0, which seemed a bit artificial. Although it would be interesting to be able to trade in google adwords.
26 March, 2007
The first part of the O’Reilly Radar Executive Briefing mainly dealt with the way manufacturing of hardware is changing. It started off with a talk about the Make: magazine with the editor and publisher Dale Dougherty. Basically what is happening is that a some people like to hack hardware to make new things. This could be a trend in the future. However, at the same time the observation was that in US education (and probably in more places) hands-on learning is happening less often. Already ‘shop-class’ where you learn how to build things disappears because of liability issues.
The session continued with Brian Warshawsky of Potenco, who showed some prototypes of a power supply for the ‘One Laptop per Child’ project. The way they are working has changed dramatically through the use of (real) rapid prototyping. With the use of contract manufacturing in China and prototyping devices like 3D printers they are able to get working prototypes of their ideas within a week.
Bunnie Huang, working on the Chumby, told a similar story. He added an open source hardware model to the picture. A common perceived problem with manufacturing in China is the potential problems with IPR. When he moved part of the manufacturing to China, the companies started offered to sign NDA’s. His reply was: “no need, all blueprints are on this URL”. In this case, the businessmodel is based on a subscription model for the widgets running on Chumby.
Simon Wardley also joined the stage and put forward the idea that hardware is become more and more malleable. Would it even go so far that it might become more malleable than software in the future? Also, this provokes the question whether if there should be ‘hardware compilers’. A compiler should be able to make decisions on how to put together products, and what materials (and manufacturing) should be chosen. The end goal of this ‘agile’ manufacturing is not to move to China, but to get manufacturing into the home, let people make their own stuff.
Of course the discussion also moved into the area of printing electronics. Not just for the sake of quick prototyping, but for the virtues of this technology. It should become possible to print (conference) bags with solar panels. One step further would be to have 3D printers that can also include the electronics into the printed models.
26 March, 2007
Something that might become handy some day. Information about RFID readers, and a python script to read them.
26 March, 2007
After a long day of travelling (Düsseldorf – Atlanta – San Diego) I arrived in San Diego saturday night. After a quick bite and a look at the view from my room, it was time to sleep some jetlag away. (It turned out that there is no need to worry about anything, there are three bibles in the bedside drawer:)
Sunday I could enjoy freely, the conference hasn’t started yet. It was a great day to see some of San Diego. I’ve walked around a lot, and saw most of the downtown area, including the Mall. In the port there was a huge aircraft carrier (now a museum), and later in the afternoon (while writing this post) a submarine came into port.
More posts on Etech 2007 later.
24 March, 2007
Interesting, this might be the first major ‘django’ company that has been bought by a big corporation. Tabblo is a really cool application, curious to see how HP will market it.