Screenshot: A Weblog

Gmail.edu

2010-02-27T01:43:14Z

I know that a lot of schools are looking at outsourcing more and more services to save money - both physical services like facilities maintenance and technological services. I liked this student perspective in a recent Yale Daily News on Yale's plan to transition their email to Gmail. Besides enumerating some of the privacy and accessibility concerns that such plans have raised, the article argues for an open process when making such a significant change. It seems, from these students' perspective at least, there are questions they would like to have answered about the services Gmail will provide before a switch is made. I cannot comment on what types of opportunities for information and feedback Yale may have provided. But taking this article at face value, these types of open conversations can be time consuming, but particularly at a college or university I think there is so much value to helping students practice being part of complex decisions where multiple factors are being weighed, that the type of transparency being called for has a strong connection to the educational goals of these institutions.

Cookie-Free Tracking

2010-02-10T19:22:57Z

I am teaching information security this term, so expect more security related content over the next couple of months. First up, if you're wondering how easily traceable you are on the internet, visit Panopticlick. A project from the Electronic Frontier Foundation, the site looks at what unique information it can collect about you via your browser - even if you have cookies off. Based on information like your time zone, your screen size, what fonts your browser has access to, and what plug-ins you have, a "fingerprint" is created for you, and the compared against the fingerprints of all of the other users to try the site so far. My browser tested as unique against the 610,703 tested at the site so far. While that doesn't prove that my configuration is 100% unique, and certainly most if not all of that information is easily changeable if you are trying to hide, it does suggest that if sites (whether corporate or government) were interested in tracking visitors and sharing information about what people are doing to build up profiles, it would be quite plausible.

Coming to the end of making even gaming painful and hard....

2010-01-28T06:10:28Z

It is a convenient coincidence that a number of interesting articles about different aspects of gaming and gaming culture have surfaced in time for the last week of my intersession course.

Develop magazine recounts the stories of the ten biggest flops as games in the past decade - the reasons behind the failures are the most interesting part
MSNBC reports that the size of certain portions of your brain predicts your ability at playing video games - an odd claim given that "video game" is a very broad category and I don't tend to think that there is a single set of skills that lead to success in all games, which is consistent with some of the details buried later in the article
In good news for my students, learning how to create computer games can improve other student skills and can be effective in broadening interest in computing
On the down side, they may find themselves at reduced risk for rickets if they play too much
One game player muses on how to compute the value of a game based on a combination of cost, enjoyment and time spent playing it - seems to me there might need to be an exponential in there somewhere
Besides describing a neat Project Natal game, 2 Finger Heroes, that is on the scrap heap, this article alludes to the problem of localization in games, in this case that gestures do not always translate across cultures
Finally, if you think the life of a game developer seems like fun and games, take a look at this open letter from wives of Rockstar employees protesting their work situation - though be aware this is just one person's presentation of the situation; there is also an interesting implied gender to the developers at Rockstar in the letter

Next Advertising Frontier

2010-01-24T15:21:28Z

There are a lot of interesting angles to the possibility that Google is developing technology to detect billboards and other ads in Google Maps street view images and replace them with their own ads. This news is based on a patent application, so it may not even happen. But it raises the question of whether Google even wants to get into doctoring their images to such a degree. Blurring out a detail or removing an image is one thing, but if users know that what they see may not be reality, will they lose faith in the reliability of the system? Having the imposed ad integrate seamlessly with the image would probably be necessary for it to be effective, but I would personally prefer to be able to identify imposed content.

There actually seem to be a set of patent applications that the articles I have seen are talking about and merging the discussion of - one that automatically finds ads in images and one that manages auctions of advertising space. The image-processing patent (application number 20100004995) is quite long so I've only skimmed it, but I would be very interested if this were pursued on how exactly an advertisement was defined and if there was an intention to include a person in the loop to ensure that only actual advertising space was being treated as such. Near me there are giant "Welcome To Pennsylvania" (or West Virginia...) signs that could look like ads but which convey map information that ought not to be lost. Simply finding large rectangular details won't do it.

This also seems to raise the possibility that billboard advertisers will start acting like spammers, carefully crafting ads that both advertise and yet are missed by Google's algorithm. Google's online actions might end up changing the way our real-world advertising looks.

Up is Down. Down is Up.

2010-01-20T03:06:09Z

I've been playing a lot of games recently (but for work!) and I'm trying to be attentive to what makes me remember a game. Being a Tetris fan got me to check out First-Person Tetris, but I expected to find it gimmicky - when you rotate, the screen rotates around the piece instead of the other way around. But it ended up being a nice variation on the original that adds a small bit of extra complexity to a familiar game. It adds a single thing that complements the game play nicely, and executes it well. Unfortunately, after a few levels nausea can set in with the rotating screen, if you are prone to that sort of thing.

No Wookies in the classroom

2009-12-23T14:38:23Z

This article via Wired about whether geeky decorations turn women away from computer science has me conflicted. The article is definitely provocatively titled, "Star Trek Stops Women From Becoming Computer Scientists", but the underlying study being reported shows that sitting in a room with Star Trek decor correlates with women responding more negatively to a survey of attitudes about computer science, with men not showing the same effect.

As always with this type of study, there are things to poke at - would other strongly themed decors have the same result? What about a non-neutral room with lots of academic science posters and pictures? Or does a decor with any geeky content correlate with negative attitudes? Is the negative impact due to association with the geeky culture and the types of people the women think of as doing those jobs, so the issue is it being Star Trek, which triggers all sorts of stereotypes? Or is there a disinclination to associate so strongly with science that it defines one's whole life including the decor of one's space, and it doesn't matter what the nature of that geeky decor is?

I obviously have ulterior motives for wondering - I inevitably end up decorating my spaces. And I'm a bit of a geek. I might even have a *very* *tiny* Enterprise model in my office that someone gave me, as well as the obligatory Escher calendar, conference posters, etc. Is this subtly hostile to female students, and do the quilts and curtains that I've hung help counteract that? Should I take down the XKCD cartoons I've hung on my lab door? I don't want to - I like that students read them and will tell me "oh, now that I've taken your AI class I actually understand that!". But I can also see how that type of imagery projects the message that you are either in the group and get it, or outside the group and don't. I would be really curious if one gets the same effect with, say, physics or math.

So I'm thinking about it.... It adds to the oddness for me that I am the only female professor in my department, and I'm definitely the one with the strongest inclination towards geeky decor. Probably with the strongest inclination towards decor in general. I would have thought personalizing a space would show an appealing warmth and personality, which one might think would have a positive impact on attracting and retaining underrepresented minorities. But maybe I ought to think about bringing my Hermione wand home...

Let Turk research for you

2009-11-02T02:47:49Z

Being interested in, and recently done a fair bit of reading about, human subjects research, this article from a couple of months ago about using Amazon's Mechanical Turk for running human subjects experiments caught my eye. It's a pretty neat idea for pulling in volunteers from a variety of demographics, with some obvious concerns - are people who they say they are, and how much effort will they put into the experiment being the major obvious ones. The article gives some nice specific suggestions for avoiding biases, cheaters, testing the correctness of your study, and addressing other issues. If done well, this seems like a clever way to recruit and manage anonymous subjects.

Best Bots

2009-10-31T04:08:34Z

I've noticed that whenever I teach a subject, the news appears to be disproportionately full of stories about that topic (and yes, I do understand that is not the case...). These are some of my favorites I have found recently:

I linked to the BigDog project several months ago - Boston Dynamics has outdone themselves, pushing their sophisticated walking abilities into a bipedal robot, Petman. The video on this is amazing! [via many places, including C.]
On the opposite end of the spectrum, iRobot has developed a "morphing blob" robot based on a structure of inflatable air pockets. [via many places, including J.]
These interconnected warehouse robots are an incredibly cool example of what modern robotics can do right now - these particular ones work together to speed up collection of items within a warehouse in preparation for shipping. Watch out for the cute "Bot Crossing" sign a minute and a half in.
A nice transition from this semester into the next - researchers at University of Washington have a nice summary of a paper about security and privacy risks with future household robots.
Finally, who doesn't want to watch kittens on a Roomba again!

Does not yet wash the bugs off my windows

2009-09-21T13:53:03Z

This is a robotics application I've never run into before, but it is actually being used in the Netherlands: a gas pumping robot. The video shown makes it look like it works pretty well too, and while it seems slow the article says it speeds things up enough to actually increase their number of sales per pump. I do wonder how much of that is the fact that, to use the pump, you preregister with your credit card number and preferred type of gas (so that you don't interact with the robot at all, you just pull up and sit while the robot does its thing). The RFiD tag alone would eliminate the time I spend fumbling around with my credit card and receipt and getting everything in and out of my wallet. It may be that the advantage of having the robot fill your tank is just enough to get people to take advantage of the RFiD technology.

Data Liberation

2009-09-18T13:59:16Z

The news from earlier in the week that Google is committing to providing users easy and free ways to move their data in and out of Google products addresses one of the concerns I have had about cloud computing. I understand the appeal in terms of expense, and I have had good experiences when I have used Google Docs for collaboration. But often the content being created is stuff that I may want around many years in the future, or also available off-line on a flash drive (we can pretend the entire world is networked, but in just the past two months I can think of three different trips that left me in locations with no convenient network connectivity).

We are even running into this with the course management systems we are looking at - not all of them offer the same degree of data portability. It seems that part of the issue there is a lack of clear "neutral" format to export all courses too, which certainly makes sense given the diversity of the tools available. But there are also systems where you can access old courses in perpetuity on their servers but cannot export the course to your own computer - where in perpetuity I fear means only for the lifetime of your school's contract with the provider. This type of lock-in makes me very nervous.

A "fair" schedule lets me sleep in until at least 7

2009-09-17T04:26:42Z

We've been talking about various types of scheduling problems in my AI class, so this local article about computer modeling used to schedule sports games caught my eye. It is an interesting constraint problem - not just the number of games, mix of who plays who, and frequency of games, but particular rules based on amount of time needed to set aside for travel and other issues of fairness. It is particularly worthwhile to think about the advantages this system offers when changes occur that make a planned upon schedule no longer acceptable. Often, the human response to that is to try to find the solution that requires the fewest shifts possible, in part because it avoids "messing up" large parts of the already-difficult-to-construct schedule. With this type of software in place, it becomes debatable whether the fix with the fewest changes is optimal compared to the fix that results in a new global optimization. This is probably a place where knowing a bit more about sports would help me.

Fashion PSA

2009-08-23T23:17:43Z

While I am willing to concede that in general, I should never, ever attempt to give fashion advice, I am still comfortable saying that nobody needs eyes painted on the back of their jeans so that their ass winks as they walk.

Don't just watch the first few seconds and stop the video - make sure you skip forward until you see the action/cut boards and the ducks at least. Based on the notes in the Consumerist article via which I found this, the eyes apparently only work if your pants are sufficiently snug. I'm surprised they stayed classy enough to not have a version that's a big pair of hands that squeeze closed on your ass every time you take a step. I honestly cannot imagine what I would do if I were to see someone in these in real life.

Same low protein, less chlorine

2009-08-22T15:39:45Z

I am unreasonably excited that King Arthur Flour is coming out with an unbleached cake flour. I love their bread flour, and my entirely-uneducated-bias is against bleached flours. I'm really hopeful that this shows up in my local food store. Sadly, I cannot even find regular cake flour in my local food store, so my hopes are not too high. Anyone who live around me have suggestions of where I can get cake flour around here? My Cake Bible has been pouting at me every time I use regular all-purpose in a birthday cake, but I cannot bring myself to order flour over the internet.

Don't let them swab you

2009-08-21T13:56:46Z

In a similar vein to yesterday's post on data, I just noticed this NYTimes story reporting that DNA evidence can be fabricated in a lab to match a target, given a sample of their DNA, or their DNA profile from an earlier sample. Even if it seems expensive and cumbersome to start fabricating DNA evidence to frame people, there are some very interesting legal implications here just from this process being known.

While the article says that they were able to fabricate blood that a testing lab processed as if it were from the target rather than the actual source, one thing it does not comment on is whether, with further or different evaluation, it would be possible to detect that manipulation was performed.

Either way, the clock has now started on the race to see which crime procedural show airs a story based around this first....

Never admit you like Star Wars

2009-08-20T20:44:29Z

I really enjoyed this article about how little information is needed to personally identify you. The answer is, a lot less than you might think, and using more benign information than you might imagine. 1990 census data showed that 87% of people in the US could be identified based on their zip code, gender and date of birth. I found their results on movie preferences even more interesting: "Knowing just a little about a subscriber--say, six to eight movie preferences, the type of thing you might post on a social-networking site--the researchers found that they could pick out your anonymous Netflix profile, if you had one in the set."

As the article goes on, bringing up RFiD tags, phone tracking, and surveillance cameras, you get the idea that the only real protection you have is hoping that nobody really wants to bother tracking so much information about you. It is not plausible to perfectly control what information you allow to be released when small, diverse facts when grouped with the right databases can pinpoint who you are. Better to be aware that this is now possible and focus on having legislation that controls how such data can be used and what individual rights we have to access our own data, correct it, and ideally remove it.