The flies 'get' the task after only about 8 minutes of training and keep clear of the punished color/turning direction even if the heat was switched off afterwards. When I analyzed what the flies have learned, the turning, the colors or both, it turned out they learned only the colors. I tested the color memory by asking them to choose colors with a different behavior. The experiment for the flies can be imagined like first learning to drive a car in the US or the European continent, only to then have to drive a car in the UK or Australia, with everything reversed. After 8 minutes of training, the flies could switch from one car to another. When I did the same analysis after 16 minutes of training, I found that both the behavior and the colors had been learned. In the transfer test, the flies now failed to choose the previously unpunished color with the novel behavior. In the car example, if you practice car driving long enough, driving in a different car in a different traffic system becomes much harder then if you've just started learning how to drive. Basically, it becomes difficult to kick the habit, even if you're a fly. These results don't only match well with vertebrate and human studies, it also dovetails nicely with our recent discovery that the behavior-learning and the world-learning (for the flies: turning and color-learning, respectively) systems can be distinguished at the molecular level.
Once we saw the interaction between behavior and world-learning, we started to use the genetic toolkit that comes with the model system Drosophila and found that a very well-studied insect brain region, the mushroom-bodies, is involved in some of the interactions underlying the regulation of habit formation in flies.
These results now allow us for the first time to tease apart the mechanisms of interacting memory systems at the molecular and the circuit level. If you think that's interesting, you can get the new paper in the download section.

As you can see, the video shows a long sequence of various mechanical and chemical cause-and-effect sequences, a chain reaction. Some of these sequences are difficult to comprehend and believe without some basic chemistry and physics. The intricacy of the design is impressive and the whole sequence does make one think about determinism in the macroscopic world. It made me reflect about my own work on spontaneous behavior, which, while not uncaused (it's caused by a biological mechanism in the animal), lacks the external sensory trigger which is commonly taken for granted.
The next piece was something for all senses: an old cement mixer, refurbished to roast popcorn:

When it was running, the machine was loud, the entire museum smelled of popcorn and if you wanted, you could eat the freshly popped corn. A very ironic way of fusing the sensations during the procedure with associations one has with the original purpose of the machine and with the smell and taste of the popcorn.
The third object was a very fragile-looking piece, seemingly cobbled together from what the artist had standing around:

It was not the first time I was really angry that the iPhone 3G can't record videos! What you can't see on this picture is that the sheets of paper are rotating against one another such that they are always pushing against each other. This arrangement elicited the most peculiar impression and was surprisingly aesthetic. We even thought we could build something like this (maybe with light) for the livingroom... beautiful.
The next piece was the one that captivated me the most. For me it embodied the essence of 'romantic machines'. It's "Time Machine" (there's also video on that page) by Robert Barta:

There is a small, red electric locomotive (which so reminded me of "the little engine that could") running on a big horizontal wheel. The wheel is rotating at such a speed, that the engine is basically at a standstill, with reference to the room. The wheel and the train are both symbols for some of humanity's greatest inventions, yet the tiny engine on the gigantic wheel feels like almost a Metropolisian allegory of insignificant beings running as fast as they can only to stay where they are. Talk about a romantic machine! The little engine also reminded me of today's scientists struggling to balance experiments, teaching, writing and conference traveling just to get from one short-term-contract to the next in this publish-or-perish world.
A very dynamic exhibit was O2 by Zilvinas Kempinas. There was a fan which kept a loop of magnetic tape (video recording tape) hovering on the wall. Whenever your hand got into the airstream, the loop changed shape, seemingly alive. This video is pretty similar:
I was mesmerized at the sight of the hovering band and felt why it fitted very well into the category of 'romantic machines'.
The next exhibit that invited me to stop and think was a mirror my Johanna Smiatek. This mirror would start to vibrate when you got too close, making it impossible to discern any details any more. This one was reminiscent of Narcissus, who fell in love with his image in a pool. If you get too close to the surface and touch it, you also destroy the image.
The technically most advanced exhibit was the bit.reflection installation by Julius Popp of bit.fall fame:

The mirrors reflect the light onto the wall and form words, pulled from news-sites on ther internet. At first, only a few letters appear, then the whole word, which quickly becomes illegible as it's transformed into the next word. Very much like bit.fall, it captures nicely the ephemeral nature of news these days. For a few media cycles, something incredibly impressive or important happens, reverberates in the various media, only to be dropped and forgotten and sometimes destroyed soon after. You can find videos here and here which show the machine working in grerat detail. The video below shows several Julius Popp pieces and starts bit-reflection at 2:57.

I've found three posts, so far which cover Google Wave from a scientist's perspective. Martin Fenner asks that Google Wave mustn't forget the scientists, Ricardo Vidal dreams of the perfect (Google) Wave to surf the streams and Cameron Neylon thinks that everything has changed in his wave of adulation. Of course, I first learned about Google Wave from the various threads on FriendFeed.
It's not without some irony that the brain child of scientists - the internet - is now developing at such a rate, that we hopelessly outpaced scientists feel the urge to plead to a private company to not abandon us poor scientists. Some of us, after all, are spending roughly 5 billion € on the most complex experiment in the history of humanity. Embarrassingly, I would posit that most of today's scientists use the internet for science at roughly the level of 1994: Browsing and e-mail. However, this should not keep us from joining the evolution of Web 3.0 and help develop Science 2.0 (see, even in our versioning we lag behind the general public). What could Google Wave do to bring the way science is done into the 21st century?

1. A great example of the abovementioned anachronism in science is paper writing. The bread-and-butter task in our publish-or-perish world. How do today's scientists write papers? They edit documents in MS Word and then send each other versions via e-mail. Which is what we have been doing since about 1994...
   There have since been a few developments which a few scientists have picked up. Social bookmarking à la Connotea or citeUlike is one area and social networks such as FriendFeed or Nature Networks is another (and many scientists use Facebook privately). One of the latest of them which I also use is Mendeley. Mendeley cooperates with citeUlike such that you can add papers from a website to your collection and then add/edit the references in your manuscript from your collection. Having had a look at the editing capabilities of Google Wave, it is clear that one of the first things we as scientists should do is to add collaborative reference management and figure/table numbering to the rich text editing capabilities. Martin Fenner lists this feature on his blog post and we're discussing it at FriendFeed. Of course, collaborative editing doesn't stop at papers: grants, lab-wikis, institute websites, manuals, protocols, fridge content, who's up for journal club?, which machine is currently broken/working, where is item X?, who's going to the conference? Which posters are you going to look at?.... the list goes on and on! Google Wave has solutions for all of these processes - today.
   Obviously, any sort of peer-review is also set to profit tremendously from such collaborative editing - isn't that one definition of peer-review?
   
2. Another workflow that is seriously outdated is the way collaborations are run during the project. Today, hardly a project is conducted by the lone researcher in a dark lab. Mostly, it's (international, transcontinental) teams of scientists with various specialisations. How do scientists today update each other on their various projects? Scientist A meets scientist B at a conference: "Oh, and by the way, we found that thing we were supposed to be looking for - it's not what you thought". Or: scientist A emails scientist B: "hey, it just occurred to me it's been like forever. Have you guys found that thing yet?". If as many scientists will be using Google Wave as are using e-mail today, the progress is posted (automatically, if the experiment is automated) to the wave of the project for all participants to see in real time. This is actually a feature of a grant proposal we just sent to the German funding agency (DFG). We used a predecessor of Google Wave, Google documents, to draft the proposal. We might as well now tell the funding agency that the methodology we said we would employ in our grant proposal is already outdated and that we'll be looking into using Google Wave. See also our discussion of this grant on FriendFeed. Automated data logging aspect was also covered in Cameron Neylon's post.
   
3. If everybody is using Google Wave as people are using e-mail today, every contribution of each scientist to every project will be logged and timestamped. Author contributions, database developments/contributions, expriments, ideas, interpretations, anything could be used and attributed. We'd have an entirely new reputation system at our hands, one that could finally replace the centuries old 'publish-or-perish'. If people are saying that e-mail died last Thursday, the Impact Factor died along with it.

A main factor driving adoption (besides the features and potential) of Google Wave will be that it's going to be Open Source as well as a new standard (federal) protocol, similar to the SMTP on which e-mail is running today. This means, you could set up your own version of Google Wave and keep it and any contents in it entirely secret and hidden from public view, just like current intranets. It will be interesting how companies like Facebook and Friendfeed will react to Google Wave, as their functionality is in serious competition by Google Wave. Some think Friendfeed might actually benefit.

In all this hype and enthusiasm, it only remains to wonder what a company could possibly gain from making more than two years of R&D open source and to push for an open standard. Surely, Google will not give all this effort away for free. They may speculate on a head start in developing Wave clients (or servers). They may release basically unusuable code in to the open. Who knows. True non-profit standards are rarely developed by for-profit industries, so scientists need to remain skeptical and prepared to find their own solutions until we have all the code in our hands.

UPDATE: There's a great article over at the Chicago Sunday Times explaining what exactly Google Wave is.