Random thoughts of a computer scientist who is working behind the enemy lines; and lately turned into a double agent.
Thursday, February 21, 2008
Search and the New Economy: Slides and Videos
I am now done with the Winter'08 offering of the class "Search and the New Economy". For those of you who are interested in looking at the material, I have started putting the material online. For now, I just have the slides and recorded lectures posted at the public class website. You can also look at the class roster and at the prediction market site of the class. I will also post the assignments and my experience on how everything worked and what I plan to change in the future.
Monday, February 18, 2008
Course Evaluations and Prediction Markets
In my graduate class "Search and the New Economy" we covered (among other topics) the concept of the "wisdom of the crowds" and the ideas behind prediction markets. As an assignment, students had to create contracts for various prediction markets and post them at the class website at http://searchandtheneweconomy.inklingmarkets.com [not functioning anymore as the pilot expired]
To get the site going, I created a market trying to predict the course evaluation. (I am always curious to find out how students like the class.) So far (as of February 18th) the class rating according to the weighted market prediction is 6.23/7.00. Such an evaluation will be a pretty good rating, if it is of course accurate :-). I am curious to see how accurate the result will be when the ctual evaluations are released in the next few weeks.
To get the site going, I created a market trying to predict the course evaluation. (I am always curious to find out how students like the class.) So far (as of February 18th) the class rating according to the weighted market prediction is 6.23/7.00. Such an evaluation will be a pretty good rating, if it is of course accurate :-). I am curious to see how accurate the result will be when the ctual evaluations are released in the next few weeks.
PREDICTIONS CLOSING VALUE EXPIRED AT
From 6.5 to 7.0 $36.86/ $0.00
From 6.0 to 6.5 $49.95/ $100.00
From 1.0 to 5.0 $4.23/ $0.00
From 5.0 to 5.5 $4.60/ $0.00
From 5.5 to 6.0 $4.34/ $0.00
The weighted prediction market was 6.23/7.00. The final evaluation received for the class was 6.3/7.0.
Wednesday, February 6, 2008
Privacy and Intellectual Property: Two Sides of the Same Coin?
I have been catching up my blog reading and I run into an interesting posting by William Cohen about personal control of personal information, where he talks about the idea of controlling the
personal information that we generate by simply "being" in an online world. He points to the new Attention Profiling Markup Language (APML) protocol, that can be used to capture and standardize the way that we record personal information, such as browsing history, video rental preferences, and so on. If we have such a standardized preference profile, then we can start building interesting applications, since we can have access to a much more holistic profile for a given user. Fernando Pereira on the other hand points out to issues related to privacy and security controls. Who controls and has access to such information?
Reading the arguments, I started thinking the relation between intellectual property and privacy. Fundamentally, both deal with control of information. Intellectual property deals with the control of data that are generated by someone and are "creative works" and privacy deals with the control of "non-creative" data that are generated by someone as a result of simply being and acting in a society.
There is a tremendous amount of attempts to enforce the control of intellectual property, with Digital Rights Management being a prominent, controversial example. Additionally, there are proposals for embeddings watermaks in songs, videos, and books to be able to know who has leaked a protected file.
Interestingly enough, I have not seen anyone proposing to use DRMs and watermaks for controlling the availability of personal information. I may want to reveal my video rental preferences to Amazon and Netflix, but I do not want other companies to have access to these data. Similarly, I trust Google with my search history, but not anyone else. Given APML-coded profiles, together with DRMs, we can specify who has access to such data, in the same way that music labels are now trying to control who can listen to a particular song.
Maybe this is the real application for DRMs, and not ebooks, music, and videos?
PS: One obstacle is, of course, the fact that the privacy-sensitive data are not controlled simply by the person who generates them. To consider some types of data to be a privacy threat, there is another party that observes the same data: what I buy and what I view are actions observed by the party that sells me something or allows me to browse through the website. Therefore, private data are a form of shared property and while one party (e.g., the user) may be able to control how these data are distributed, the other party (e.g., Google) gets the same data in a format that is beyond the control of others. This makes privacy-related data to be a little bit more complicated to deal with compared to data that fall under the provisions of the intellectual property law.
personal information that we generate by simply "being" in an online world. He points to the new Attention Profiling Markup Language (APML) protocol, that can be used to capture and standardize the way that we record personal information, such as browsing history, video rental preferences, and so on. If we have such a standardized preference profile, then we can start building interesting applications, since we can have access to a much more holistic profile for a given user. Fernando Pereira on the other hand points out to issues related to privacy and security controls. Who controls and has access to such information?
Reading the arguments, I started thinking the relation between intellectual property and privacy. Fundamentally, both deal with control of information. Intellectual property deals with the control of data that are generated by someone and are "creative works" and privacy deals with the control of "non-creative" data that are generated by someone as a result of simply being and acting in a society.
There is a tremendous amount of attempts to enforce the control of intellectual property, with Digital Rights Management being a prominent, controversial example. Additionally, there are proposals for embeddings watermaks in songs, videos, and books to be able to know who has leaked a protected file.
Interestingly enough, I have not seen anyone proposing to use DRMs and watermaks for controlling the availability of personal information. I may want to reveal my video rental preferences to Amazon and Netflix, but I do not want other companies to have access to these data. Similarly, I trust Google with my search history, but not anyone else. Given APML-coded profiles, together with DRMs, we can specify who has access to such data, in the same way that music labels are now trying to control who can listen to a particular song.
Maybe this is the real application for DRMs, and not ebooks, music, and videos?
PS: One obstacle is, of course, the fact that the privacy-sensitive data are not controlled simply by the person who generates them. To consider some types of data to be a privacy threat, there is another party that observes the same data: what I buy and what I view are actions observed by the party that sells me something or allows me to browse through the website. Therefore, private data are a form of shared property and while one party (e.g., the user) may be able to control how these data are distributed, the other party (e.g., Google) gets the same data in a format that is beyond the control of others. This makes privacy-related data to be a little bit more complicated to deal with compared to data that fall under the provisions of the intellectual property law.
Tuesday, February 5, 2008
Mechanical Turk, Charity, and Incentives
We were discussing in class today how to provide the appropriate incentives for users to participate in systems that rely on the "wisdom of the crowds". We have seen that small monetary incentives are not really a good motivation for people to join, but fun, or a sense of contribution are better sources of motivation.
Then, the topic of Mechanical Turk came up, and there was some discussion on whether the existing monetary incentives are the best possible. One idea, by Amichai Lesser, was to allow Turkers to send their rewards directly to a charity, instead of accumulating insignificant amounts of money. Given that a large number of people on Mechanical Turk complete tasks for fun and not for money, the idea of allowing users to convert their participation into direct contributions to a charity seemed like a no-brainer.
Amazon, will you implement this? :-)
Then, the topic of Mechanical Turk came up, and there was some discussion on whether the existing monetary incentives are the best possible. One idea, by Amichai Lesser, was to allow Turkers to send their rewards directly to a charity, instead of accumulating insignificant amounts of money. Given that a large number of people on Mechanical Turk complete tasks for fun and not for money, the idea of allowing users to convert their participation into direct contributions to a charity seemed like a no-brainer.
Amazon, will you implement this? :-)
Monday, February 4, 2008
How to Embed a YouTube Video in PowerPoint 2007
I wanted to show a YouTube video in my class. The easy way to do this, is to save the URL of the YouTube video, click on it, start the browser and watch the video.
However, I wanted to make it slightly more elegant and embed the video directly into the slide.
The first thing that we need to do, is to enable the insertion of Flash objects in PowerPoint. The instructions are here, but I repeat them below
Just in case you were wondering why I wanted this particular video in my class: This is part of a viral campaign built by Honda for the Honda Element car, back in 2006. The goal was to make the Honda Element car relevant to keywords that are inexpensive to bid on. (In this case "funny commercials" that had a cost of $0.10 or so.) The target keywords were queries submitted frequently by users in the target demographic (young males, just graduated from college), so that Honda could attract attention to the new car, without competing for keywords like "car," "honda," "suv" that were going for $2-$3 per click. Great idea, and great commercials. Take a look at the ad below and click on the related ads to see more ads from the campaign.
However, I wanted to make it slightly more elegant and embed the video directly into the slide.
The first thing that we need to do, is to enable the insertion of Flash objects in PowerPoint. The instructions are here, but I repeat them below
- Open PowerPoint and create a slide.
- Click the Microsoft Office Button , and then click PowerPoint Options.
- Click Popular, and then under Top options for working with PowerPoint, select the Show Developer tab in the Ribbon check box, and then click OK.
- On the Developer tab, in the Controls group, click More Controls .
- In the (long!) list of controls, click Shockwave Flash Object, click OK, and then drag on the slide to draw the control.
- Right-click the Shockwave Flash Object, and then click Properties.
- On the Alphabetic tab, click the Movie property.
- In the value column enter the URL of the Flash object, which you can get as follows:
- Take the canonical form of a YouTube URL, e.g., http://www.youtube.com/watch?v=QBX0Jfr62Kc
- Transform the URL into http://www.youtube.com/v/QBX0Jfr62Kc
- Enter this URL as the value of the Movie property.
Just in case you were wondering why I wanted this particular video in my class: This is part of a viral campaign built by Honda for the Honda Element car, back in 2006. The goal was to make the Honda Element car relevant to keywords that are inexpensive to bid on. (In this case "funny commercials" that had a cost of $0.10 or so.) The target keywords were queries submitted frequently by users in the target demographic (young males, just graduated from college), so that Honda could attract attention to the new car, without competing for keywords like "car," "honda," "suv" that were going for $2-$3 per click. Great idea, and great commercials. Take a look at the ad below and click on the related ads to see more ads from the campaign.
Sunday, February 3, 2008
Google-ridiculous
Is it only me that considers ridiculous the reaction of Google to the bid of Microsoft for Yahoo! ?
I understand that Google would like to put as many obstacles to such a deal as possible. But claiming that the deal is anticompetitive because of the "overwhelming share of instant messaging and web email accounts" of the resulting entity?
Come on...
I understand that Google would like to put as many obstacles to such a deal as possible. But claiming that the deal is anticompetitive because of the "overwhelming share of instant messaging and web email accounts" of the resulting entity?
Come on...
Subscribe to:
Posts (Atom)