Windows 8 Start Screen

This blog post is technology specific, but has elements of how humans may use spatial arrangements, color, and grouping to find applications in the changed start screen in Windows 8.

Personalization has its dark side. Just like other information access technologies..

This is my reflection on a recent TED talk that I recommend to watch:

http://www.ted.com/talks/eli_pariser_beware_online_filter_bubbles.html

Eli Pariser observed that information personalization is everywhere and it creates a bubble that shields one from the real world. What a surprise!

I hope that we all understand that some kind of tunnel vision is a back side of any information access technology that attempts to help one funding a needle in a haystack. Card catalog and simple (non-adaptive) search were steps in this direction. When you search in a classic library catalog for books you only see books relevant to your query. It's already a form of tunnel vision. Without library catalog search functionality you would have to browse the shelves and have chances to encounter some nice books you were not aware about. Wouldn't it be nice to keep finding book this way? But alas, we all use search instead of digging into the library stacks. Search shields you from mostly irrelevant stuff it and make your work more efficient. Loosing some gems on the way, but making many new things possible.

Nice to see that people like Eli finally can recognize that personalization makes another step into this direction. This is a sign that recommender systems area is coming of age and going through the same process as user modeling went about 10 years ago - recognizing problems associated with personalization. In user modeling area the problem of one-shot uncontrolled personalization has been discussed at length and a number of suggestions of controllable, transparent personalization was made. I guess, time to focus on that for recommender systems. Recognizing back side of every technology is vital to make it better!

Redefining the goals of a PhD program/degree

I found the opinions expressed in this article extremely interesting and true. The extreme specialization offered from PhD systems today makes their outcome interesting and relevant to a very small part of the general crowd. I truly believe that in iSchools the barriers across disciplines are indeed less strict but maybe we need to put even more effort. Definitely there are many challenges that we need to account for, for the reformation discussed in the article, but this article is a start to stimulate more thoughts and to motivate taking actions towards this direction.

Let's pick our ballot at random next time!

In this interesting article, it is shown that a parliament would govern a society more efficient if some of the legislators were elected "randomly". These randomly selected legislators are not politicians but just civilians (that most probably satisfy a set of criteria). The paper provides a golden rule for the number of independent legislators taking into account the size of the parliament as well as the elected percentage of each party in the parliament.

It is surprising that adding randomness in a social system boosts its efficiency. In the same direction it will be interesting to see if adding randomness in an online social system/network helps it achieve its goal. E.g., for twitter, if we randomly follow some people does this increase the entropy of the information we obtain from the "tweets" ? Online social networks suffer from a large number of issues (e.g., spam, multiple accounts, etc.) that add challenges to similar studies and/or approaches to boost the efficiency of the underlying system.

Wikipedia vs. the Universal Decimal Classification

This is an interesting graphic comparing Wikipedia's social classification and the Universal Decimal Classification.