Dagstuhl 15051 Notes day 5

The last, fifth, day we showed each other what we did the day before, and we discussed the future.






Notes from the ending discussion about the future

Dagstuhl 15051 Day4 - Contrabot and Shrubbery games

On the fourth day we made games!

They are mentioned here:
AI-Based Games: Contrabot and What Did You Do? Cook, M., Eladhari, M.P., Smith, A., Smith, G., Thompson, T., Togelius, J. and Zook, A. Playable Demo Track, Proceedings of the 2015 Conference on the Foundations of Digital Games (FDG 2015). Monterey, CA, June 22-25, 2015.

Dagstuhl 15051 Day 3- Photo Diary and notes

Blog about the third day of Dagstuhl seminar 15051 where I discussed agents in fiction-settings,  considered pros and cons of answer set programming vs genetic algoritms, and discovered that my sweater is actually a brush holder. Also flopped around in mud.


On the third day we continued in our groups. We narrowed down the discussion on believable agents to concern contexts of dramatic settings - when they are actors or actants in fiction. It was a good discussion.  

There are some notes from the discussion below, but I'm not sure if they at all make sense to someone who was not there. It was a great group - it is not everyday one has the chance to talk narrative with some of the most prominent researchers in interactive narrative and narrative generation.

I noticed that that Michael Young had interesting decorations on his socks,  little friendly fluff-knobs. He said he had put them on inside out. Nevertheless, they added some cheer. Brian Magerko presented our work to the group.






While taking notes, I discovered that my sweater was perfect for storing brushes in. I made it easier. I also had the handbag that has a magnetic lock, so the little tin with my colour pans stick to it. (I don't need to worry about dropping the colour tin on the floor and start fussing around after it.)

After the presentations we went for a long walk in the woods. We flopped around in a mix of mud and snow, and had a terrific time. 

Dagstuhl 15051 Day 2 - photo diary and notes

 On the second day we continued working in our groups. Adam showed the first prototype, a pattern matching thing - he and parts of the group had been coding it the evening before while the rest of us had strolled about in oblivion (likely the game room.)

We continued working - consolidating the notes on the day1 discussion on affordances in terms of AI capacities and game design patterns. We looked at it from a player activity pow for a while, and we prepared presentation slides, and then we plotted the next prototype.

After lunch, all groups presented the work done. I found the presentation about research metods particularly interesting, discussing how we can improve the methods so that we can make a wider range of research work possible.




Here is a watercolour note from the presentation-session.


After the coffee break we divided into new groups. I joined a group concerned with believability of agents. At this point we were a bit tired, and the group was large - 10 persons - which led to the inevitable meta-discussion of how to narrow down the subject.

Dagstuhl 15051 Day 1 Photo diary & notes

 We were cordially welcomed by Julian Togelius, Mike Preuss, Michael Mateas, Simon Lucas and Pieter Spronck (not visible in the picture), and then we had some mini-talks, introducing potential topics for the seminar. They are listed in the watercolour note below:
 



 We had a break for lunch, and then we had a couple of more pitches for topics, here is the list: We then divided into workgroups. I was curios to learn more about automated game design, so I joined groups with Dan Ashlock, Simon Colton, and Mark van Krevold. Pier Luca and Moshe Sipper joined us when they found us strategically placed close to the coffee machine.


Notes:
 


At coffee time I realised that we would be in the first groups for longer than I had first anticipated, so I jumped over to the group talking about game design for artificial intelligence - to talk about this was one of the reasons I went to the seminar in the first place.



We identified game design patterns that mesh particularly well with different AI approaches, and commenced some implementations of examples. Notes:

Dagstuhl 15051 Day 0 Photo diary - Preparation

Schloss Dagstuhl is a castle which is owned by Leibniz Center for Informatics where seminars that can advance computer science are organised. Last Sunday I travelled to seminar nr 15051 - "Artificial and Computational Intelligence in Games: Integration". [Seminar 15051 homepage]
The seminar was a follow-up of a seminar from 2012; here is a link to my post on that seminar.





I packed light to make sure checked in luggage wouldn't delay me getting on the train from Frankfurt airport to St Wendel. The train is labelled local despite that the train ride is two hours, and runs from platform 3.

On the flight I made some notes to mentally prepare for the seminar.