I don't have too much to say about this talk, as it was mostly Tufte recounting his life as an academic all the way from his undergraduate education at Stanford to posts at both Yale and Princeton, followed by a slideshow of his more recent sculpture work. There was one very fun moment for us GG followers:

Edward Tufte, master of the display of statistical information, has put up a slide with sparklines and Galileo's description of Saturn. Suddenly, he gasps -- the projection screen is too far away! How will he ever properly point to Galileo's illustration of Saturn's rings?!?!?

Never fear, GadgetGuy is here -- sitting in the third row, actually, taking notes on his MacBook. GadgetGuy quickly reaches into his pocket and produces his trusty green laser pointer. The talk is saved!

Douglas Hofstadter sponsored by Stanford Humanities Center

Most of my few run-ins with Douglas Hofstadter haved corresponded well with whatever I'm reading at the time. I read Godel, Escher, Bach during college, which connected the dots between all the different computer science classes I was taking, a feat that my professors were not interested in accomplishing. Shortly thereafter, I saw Hofstadter speak about translating Russian literature. This did not do as good of a job unifying my education.

Tonight's talk, Analogy as the Core of Cognition, was also outside the computer science domain, but my pop-sci interest in brain books has given me more dots to connect. Less than an hour after Hofstadter's talk, I read this passage in Birth of the Mind (p. 138) that struck me as almost being planted by Hofstadter for me to read:

Another critical factor may be the almost magical ability of humans to combine simple elements into more complex ones that can in turn serve as elements in futher combinations, as idea sometimes referred to as "recursion." If you can think about a ball, you can think about a big ball, and if you can think about big ball, you can think about a big ball with stripes, a big ball with stripes that lies on the beach, and so forth.

Although penned by a different author, this passage in many ways is the central idea to Hofstader's talk, which you can read more about in the extended entry. Hofstadter came out old-school with the overhead transparencies and in some spots in the notes I've used photos of his slides instead of textual transcriptions.

Beyond Menus and Toolbars in Microsoft Office
Jensen Harris, MS Office UE Team
http://blogs.msdn.com/jensenh/
BayCHI

Office 12 is an upgrade I wouldn't mind paying for, that is, assuming that work didn't let us get free copies. Those are big words for me, considering my "I Hate Microsoft" series of blog entries. I can imagine making documents faster with Office 12, or at least I can imagine making better looking documents in the same amount of time. Excel, which has been less functional for me than the spreadsheet program I used on my Apple IIe, looks like it will be come a useful tool for data analysis.

I felt that Harris made a convincing case as to why they are doing the UI revamp, that it's more than a marketing trick. The conventional wisdom out there is "Everything I need was in office [95, 97, 2000]." (For me it was Word 95). They collected a ton of data (including over a billion Office sessions) that told a different story. On a list of the top ten most requested features for Office: four of them were already in Office. Their conclusion was that "Office is good enough in that people have made peace with it." Another observation from the data was that the average user spends more time with Office (2.6 hrs/day) than they do with their spouse (2.4 hrs/day). When you take 400 million users * 2.6 hrs/day, it seems worth improving that experience. Harris rhetorically asked, "Have we reached the pinnacle of software that people use to get their job done everyday?"

It was also clear from the evolution of the Office design that a revamp was in order. The number of toolbars and taskpanes was getting out of control: they were create user interfaces to manage their user interface. There simply wasn't a good menu+toolbar paradigm for managing programs with 1500+ commands.

My notes are in the extended. I transcribed a lot of the new Office 12 elements, but text doesn't do a very good job of describing user interface. Overall, it's a more visually oriented interface: large gallery icons show what a command does and hovering over the icon gives you a preview of the result, e.g. if you hover over a font, your select is shown with that font. The UI is also much more contextualized. What you think of being on your right-click menu is now the focus of your toolbar menu. One of the most important design constraints introduced is that the UI is confined to a fix region of the screen and doesn't do any of the silly auto-hiding or auto-rearranging of Office past.

If you're really curious about the Office 12 UI I recommend visiting Harris' blog. You will find out that yes, they fixed the stupid start presentation button in Powerpoint. Also, new fonts!. Death to Times New Roman!

Summary tidbits: * "Is it worth taking this pixel away from the user?" * You must remove to simplify * Is there a classic mode? No. * They were trying to figure out what some of the commands in Excel did. Found out some options in Excel weren't hooked up into any code: "They didn't do anything!" * On Office 2000's 'auto' features: "The computer looks at the things you use the most and moves them around all the time." * On Office 2002 task panes: "New features aren't being invented, but I bet they will if we create a whole new rectangle"

Chris Jones from iRobot gave a talk at SRI. He focused mostly on iRobot's government/industrial robots (the ones with big treads that you can throw through a window) rather than the delicate Roomba and Scooba home appliance robots (FYI: the Scooba was designed to cleanup dried peanut butter from your kitchen floor in one pass).

The main line of iRobot's government/industrial robots is the packbot. I saw one of these at Robonexus awhile back -- they had it continuosly going up a staircase and dropping several feet to the ground. These things are tough (rated to 400Gs) and can handle all sorts of terrain with their tread and flipper design. The idea is that a soldier would throw this robot (e.g. through a window, around a corner, etc...) and then use a laptop to guide the robot around and get back video. You can outfit the packbot with an arm that can hold a camera or 'disruptor' for destroying explosive devices. They've even put a parachute and fan on a packbot to make it fly. A packbot is rugged enough that when it reaches the deployment zone it can just cut its parachute in order to land, or, in the case of one video he showed us, when the packbot gets piloted into a tree.

Some tidbits from the talk:

• Your home is a dangerous place: the algorithm that the Roomba uses to figure out where to clean is adapted from a minefield coverage algorithm.
• Leave no robot behind: On April 8, 2004, Packbot 129 became the first packbot to be 'killed in action.' US soldiers managed to retrieve all of its parts and it is now framed for display.
• Lest they take over: The military doesn't like hearing about robot 'autonomy,' so iRobot markets their robots to the military as being like remote control cars. Now that the military has been using them in combat operations, they are now asking for more autonomous features like "come home so we can get the hell outta here."

I went to go see Sebastian Thrun speak at Stanford about his team's winning effort in the DARPA Grand Challenge. Thrun described the contest as how to stay on the road for a very long time. It was not a general path-finding problem: DARPA gives you the route with a corridor you have to follow, as well as speed limits you have to observe for various parts of the course. Of course, DARPA didn't give perfect data. He showed a video generated from the data of Stanley driving through the most dangerous part of the course: the switchbacks of Beer Bottle Pass with a cliff on one side. DARPA's corridor was overlaid on top of Stanley's sensor data and it was easy to see that much of DARPA's corridor was actually over the cliff.

During training many traffic cones were "frequent victims of computer glitches," but Team Stanley was called "Team Boring" by the cleanup crew for their lack of incidents. The actual challenge was described as getting the data at 4AM, getting Stanley to the starting line at 6AM, and then sitting around drinking beer for several hours. The big moment came when Stanley passed CMU's Highlander as the CMU and Stanford teams listened to race radio. Thrun narrated the exciting finish for us: "[The head of DARPA] is waving his flag as if the car could see it."

Thrun said that they won mostly through luck given how close four of the teams finished. The speed limits set by DARPA for the various parts of the course were too conservative, so the cars were running below their full potential. DARPA also decided to make the course fairly easy. Asked if CMU would have won had they not had engine problems, Thrun answered, "In all likelihood, yes." Also, Team ENSCO had a faster average course time but flatted on "something really big CMU left behind" (the CMU part may have been a joke). Thrun felt that Stanford had better software than CMU and on a tougher course Stanford would have the advantage.

In the future, Thrun wants to try driving 65mph on 280, parking in a garage, convoy driving, and driving assist. Part of his motivation is to reduce traffic deaths, which a driving assist system could help prevent. He also feels that a fully automated system would change society by allowing you to use your commute time productively -- you could even drive to your destination, get out, and then send your car to go park in a parking garage farther away. These are still looking far ahead. In response to someone asking what it would take to drive at human-controlled speeds, Thrun related it to asking the Wright brothers, "If you want to fly over the Atlantic, what's missing?"

K Rajan, NASA

Talk at SRI about the introduction of AI technologies into NASA technologies, and, more specifically, the used of the MAPGEN planner for the Mars Exploration Rovers.

Although the talk was not about work practice, it was amusing for me to hear Rajan speak as one of the converted. Despite being AI technologists, they had to learn importance of human factors/understanding how people will use their technology. If I understood him correctly, it sounds like this was his first experience with social scientists on a mission (and a succesful experience at that).

The extended notes focus on the planning technology and their challenges in getting it into the MER mission. They weren't seen as a critical technology, and it wasn't until they did a 'bakeoff' competition between their technology and the current practice that they finally made it in for good.

Peter Norvig, Google; Ken Norton, Yahoo!; Mark Fletcher, Bloglines/Ask Jeeves; Udi Manber, A9; Jakob Nielsen, NN Group

I went with bp and Neil to a BayCHI talk on "Recent Innovations in Search." I agree with bp's sentiment -- there were some interesting moments, but the talk was short on revelations or insights. I guess that is to be expected as the title of the talk is past focused ("Recent Innovations") rather than future focused ("Future Innovations"); it's hard to believe that the panelists would give away yet unrevealed technologies they were working on. I'm going to try and save as much effort as possible, given that bp posted his notes. In fact, as I am going to crib from his notes, or just omit what he already has, you should just go read them instead.

d and I managed to sneak into an Ando talk at Berkeley, tiptoeing in through the sidedoor and sitting on the floor when the lights went out because all seats were gone over an hour before the talk started. I am thankful for the location of the Men's Room at Dwinelle Hall; I might not have noticed the unguarded entrance otherwise (easy to spot, given that ten-or-so people were already waiting there to sneak in).

photo: Ando's Aomori Contemporary Art Center. Photo by kwc

I really enjoy Tadao Ando's work. I'm not a fan of his most noted signature element -- concrete -- but I love the simplicity of his forms and the ways in which his buildings play with light. This talk gave a fuller survey of some of his works over the past decades, and also gave a lot of insight into his amusingly persistent mentality that guides his projects.

My notes are in the extended entry. There are a lot of large photos of his works that I've culled of the Internet to go with some of the talk notes, so the notes may load very slowly. It took a little longer than normal to put these notes together, but it was worth it, as I now have my own mini-Ando book to browse through and reflect upon.

Revolutionizing Consumer Electronics: Welcome to the TiVolution�! Paul Newby, Director of Consumer Design Margret Schmidt, Director of User Experience (UE), TiVo

I went to the TiVo talk at BayCHI/PARC. The best part of the night, perhaps, was that I have a bunch of great TiVo schwag: a TiVo doll and two new TiVo remotes -- one to replace honeyfield's remote, which has been mistaken for bunny food, and one to solve the problem we had last week of, when you lose the TiVo remote, there's no way for you to watch TV. The second best part of the night is that I learned a new TiVo feature that didn't exist on the Series 1 remote: if you press advance (the ->| button) in a list, it will jump to the end (very useful for Home Media Option).

I have detailed notes, but it's hard for me to put the effort into transcribing all of them, mainly because I've heard most of what she's said having worked at PARC for two years (big human-computer interaction focus) and having owned a TiVo for two years. As metamanda put it when I asked her if I should read Don Norman's Design of Everyday Things, she said it was good, but I've already heard everything in it multiple times. Seeing as Norman's book is somewhat of a bible for the TiVo User Experience team, I think the same applies here.

It's also hard for me to transcribe my notes because much of what was said has already been said in this interview Schmidt did for PVRBlog

There was an interesting semi-anecdote on TiVo's "overshoot correction" feature (where it jumps back a little after a fast forward). Many people think that TiVo is actually "learning" this (even across multiple users), i.e. when they fast forward and it doesn't jump to the spot that they wanted, they assume it was because they must have deviated from their normal reaction time (it's actually a hardcoded number based on the fast forward speed, derived from research).

My last thought before this switches into notes is that I wonder if TiVo is going to put an Apple-style clickwheel on the remote to replace the direction pad. The problem with navigating long lists was mentioned multiple times by them, and Margret did even mention a scrollwheel as a possibility, and it seems to me that the newest clickwheel comes the closest to carrying the TiVo direction pad concept forward.

Most of the talk isn't relevant here, so I'm only including the brief background notes, mainly because I had not heard of utile memory before.

Manuel Castells Talk at Berkeley

My notes on this talk are fairly poor. Much of what he talked about echoed what I am currently reading in his works, so skipped transcribing a lot of that.

Update: I'm enhancing my notes with those taken by peterme (he did a better job of transcribing Castell's actual phrasing). I've gotta run, but hopefully I'll get a final set of notes before my brain crashes.

Peter Norvig, Google

I don't expect anyone other than myself to actually go through all the talk notes I've been posting, but I will point to three in particular that I enjoyed and thought were useful. The first was a tutorial on Automating the Design of Visualizations, which was a great blend of cognitive science, computer science, and user studies to help try and foster better computer-generated visual designs (target audience: people with Tufte on their shelves).

Two of the invited talks, AI and the New Exploration Vision (NASA) and If Not Turing's Test, Then What? are high-level enough to be approachable. The NASA talked a lot about the technologies they are using/will use/want to use in their Mars and future lunar missions. The Turing talk discusses some of the grand challenges for AI and was also a meta talk about the properties of grand challenges.

I find myself drinking from the fire hose here -- many of the talks were topics that I knew little or nothing about, nor had any mathematical background for. If you don't understand my notes, rest assured that I may not understand them little, though being at the conference is rather like learning a foreign language in a foreign country, with the immersion making for quick study.

Update: Added to photo entries
- Mars Personal Rover
- Maze of Carnage

Multi-robot task allocation

Paul Cohen, USC ISI

This was a nice review of the pros/cons of the Turing Test as well as current grand challenges in AI, with lots of Daniel Dennett quotes.

Craig Boutilier, University of Toronto Tuomas Sandholm, CMU Rob Shields, CombineNet, Inc.

I like combinatorial auctions as they blend two fields I'm interested in. You get a Computer Science optimization problem (Knapsack problem) combined with Economic auctions.

Jie Yin

A lot of stuff about Dynamic Bayesian Networks that I don't really have a background in yet, so my notes are sparse.

Ya'akov Gal, Avid Pfeffer, Barbara Grosz'

A mix of social agents and game theory (social vs. analytic strategies)

A Case Study from Artificial Life Matthias Scheutz, Notre Dame

Peter Jarvis, Teresa Lunt, Karen Myers, NASA/PARC/SRI

Michael Montemerlo and Sebastian Thrun, Stanford

Elegant solution for managing short-range and long-range mapping data for robot navigation.

Brian Magerko, John E. Laird, Mazin Assanie, Alex Kerfoot, Devvan Stokes

Information on a storytelling environment built in Unreal Tournament.

Sebastian Thrun, Stanford

This talk was mostly over my head in terms of the math, but the work is interesting.

Dan Clancy, NASA

I enjoyed this talk -- it was a survey of NASA's current AI-based missions, including current and future Mars missions (Sojourner/Spirit/Opportunity). metamanda would have liked at least one point the talk made, which was that NASA is working on a personal rover robot to create/inspire kids, and in particular, girls. They have found in their exhibits that robots are more engaging to girls than boys, who enjoy the embodied interaction, so they see in it an opportunity to bridge a gender gap as well as inspire a future generation in NASA's vision. It was interesting how the Personal Exploration Rover pictures really did look like baby versions of the Spirit/Opportunity rovers, i.e. there was a certain amount of anthropomorphism to the vehicle, and it appeared child-like that could help engender a care-taker relationship between a kid and the robot.

Read on for notes.

Maneesh Agrawala, Julie Heiser, and Barbara Tversky Tutorial session at AAAI

Two implemented systems explored for automated design of visualizations: map routes and assembly instructions. Map routes system (LineDrive) used by MapBlast (now mappoint.msn.com).

Three parts to talk: cog sci/CS background, map routes, assembly instructions.

Tutorial session at AAAI

slides

Clear the building: Pursuit-evasion with limited field of view
Brian Gerkey
http://robotics.stanford.edu/~gerkey/
Stanford Robotics Laboratory
(with Sebastian Thrun)

Note: most of the notes are copied from the bullets on the presentation slides as the presentation was going very quickly.

Eugene Kim
Blue Oxen Associates

Kim popularized Engelbart's notion of purple numbers; purple numbers are a scheme for granular addressing of information. Purple-numbered versions of Cory Doctorow's and Lessig's newest books are online; the numbers allow you to create a link to any paragraph in the book.

The talk was much broader that just a discussion of purple numbering. It was a discussion of the broader notion of making software more interoperable with regards to linking and the inclusion of information from various sources. Kim also briefly touched upon trying to work on a shared modelling language for data that would facilitate this linking, as well as provide benefits in being able to make more intelligent uses of data.

My notes in the extended entry are fairly poor, as I wasn't too copious and the actual presentation was brief. I haven't read Kim's A Manifesto for Collaborative Tools yet, but that would probably be a better followup.

I was part of a four person panel that gave a talk on research cultures at non-PARC research center. bp posted his notes, where it should be obvious that I had very little to say. The questions focused a lot more on high-level/process questions (e.g. "when do you decide to terminate a project," "what about outsourcing research development to India"). I've only been at SRI for five months working on a single project, which probably didn't make me a very good panel member to answer questions about the research project lifecycle. I was hoping for more "culture" questions so I wouldn't sound so mute. At the very least I had a good time seeing everyone again.

I was really excited to go see Tufte today for his day long course. His writings and teachings are excellent, and I find them useful whenever I am displaying visual information, even if I cannot live up to the standards that they profess. Sometimes I wish Tufte would sell software with his design principles built-in, rather than the pie-chart glory of Excel. In fact, I learned during the talk from Peter Norvig (slightly more on this later) that the Autocomplete Wizard in Powerpoint started off as a joke by the engineering people to the marketing people, along the lines of "oh yeah, and we can just have the application fill in all the content for you." Clearly, engineers don't understand marketing.

So, on to the talk, which I discuss my rants and raves in the extended entry.

metamanda's friend Vijay gave a talk at PARC yesterday about some of his work at ILM. He worked on the movie The Hulk and is currently working on The Day After Tomorrow. I had been trying the schedule this talk while I was still with PARC, but of course he waits until I switch jobs before he comes and speaks :).

Cory Doctorow (of BoingBoing and Amazon.com: Down and Out... fame) will be speaking at BayCHI at PARC, so I thought I'd post a talk pointer:
- Tuesday, August 12, 2003: Monthly Program (BayCHI)

I didn't like this talk enough to actually transcribe the notes. Luckily, heerforce has already transcribed his notes/outline, and they're probably better than I would have done anyway.

In short, I expected a lot more from the chief scientist at Altavista. His talk was so vague compared to Eric Schmidt's Google Forum - most of the stuff Pedersen put up was general knowledge. Also, if you notice from heer's notes, his "Future" section of his talk was very vague (though he was rushed).

Bram Cohen, the creator and maintainer of BitTorrent gave a talk to our lab today. The most interesting (and surprising) aspect of his talk, IMHO, was that he based BitTorrent on the Prisoner's Dilemma. Cohen's approach was to assume that each client would act greedily instead of cooperatively, and designed the system along those principles. Although there are some slight areas in which gaming the system is possible, overall Cohen's tit-for-tat system seems to do very well and enforcing good behavior. It's nice to see game theory applied in places other than economics and politics.

I'm suprised that more commercial companies aren't using BitTorrent, especially given how expensive it can be to maintain a large Internet pipe, though some have claimed that a commercial company could not ask it's customers to use a protocol that steals their upstream bandwidth (i.e. customer's expect the companies to foot the bandwidth bill, not them). I could still imagine companies being at least supportive of BitTorrent, however, by posting clearly marked "BitTorrent" links. I could have used it yesterday when I was trying to fetch Gollum's MTV acceptance speech off the Net.

Notes on the last of the three talks giving by the Future Applications Lab folks. The last talk was on Context-Aware Photography, which seems like an interesting idea. I, too, have been bothered by the fact that digital cameras for the most part do and work exactly like their analog counterparts. Unlike them, however, they actually have some ideas of some neat things to do. They are still in a very preliminary stage.

The second talk given by the Future Applications Lab people was on Informative art .

Basically, they wanted to see if they could make abstract art serve as an informational display. There were several prototypes:
- Andy Warhol-inspired soup can countdown clock: a grid of soup cans change from one type to another indicating a countdown towards a deadline (pdf)
- Bridget Riley-inspired motion painting: vertical bars in different colors represent motion sensor input (