Synthesizing…

This is a difficult but useful assignment given Molly Wright Steeson, one of my thesis adviser, to audit the latest position I’m in with my thesis research. The requirement is to list everything I did regarding the thesis on the index cards and to write a question each of them asks in relate to one single message at the very heart of the project. The message in this case being: Designing a social network for the future that affords the creation of the impossible, the improvable, the plausible and the invisible user identity.

The cards. Sorted by research method.

The cards. Grouped by threads of topic

The thesis. Mapped with the threads.

Tale-telling UIs

Remember the ‘search stories’ by Google? It’s so interesting that search queries are actually a tale-telling media. Here I explored a similar method that ‘by seeing what shows on one’s screen you experience their mental status and the hidden stories.’

“Bits of Me” here, “Bits of Me” there

This is a study of profile management across various social medias, that involves the collection of the text data(except for their names) from 15 on-line profiles (3 for each subject) and an interpretation of the data from strangers through the portraits they draw of the person based on the profile. With the human “algorithm” on one hand, the data was also parsed through LIWC (a linguistic analysis model) that gives a personality analysis of the text on the other hand.

Interestingly, there is an overall precise match of the disposition between the interpretive drawing and the personality analysis result. And when the data is not enough to make a judgment, people were confused just as the computer does. The two methods also support each other for details they are unable to find out respectively. I found that human drawings tend to make a better guess about the person’s age, and their disposition seems to be based on the looking of some other humans that shares the same features. It is also interesting that when the linguistic analysis shows very low “negative emotions” from the text, the number of positive vs. melancholy looking faces in the drawings are almost even.

Launch the analysis process

So what exactly does it say?

The initial intention of the project is to testify my hypothesis that people use social media to strategically present themselves. And the use of one media is complementary and influenced by the rest.

Most of the persons I studies have a high awareness of “what in where”, that is separating the information about themselves in different categories and share them via different networks, and also “how to put”, that is organizing how the same information is stated differently on various media.

There is only one person who has told me before that her Facebook self barely represent her in reality, whose result has shown a consistent pattern throughout all of her networks. In that case, she has done a good job keeping the actual self away from the medias, while still keeping her image consistent.

How to both benefit and hide away from on-line search engine

Everyone has their own little strategies of using on-line search engine to find an answer. We know that tweaking the spelling or the order of the keywords can help us gain faster and more precise answer with Google search. And we more or less have a sense of the patterns of the search tools, though we don’t know exactly the algorithms they are based upon. But how about hiding from it? Using the search engine to figure out where you can hide and make the answer “difficult for other to find” is totally another type of intelligence.

To learn about how people deal with the issue of using on-line search as a “double folded weapon”, benefiting oneself and incapacitating others, I designed a game where the participants used on-line searches to both raise questions to and answer questions from their opponents.

This is how it works

It was found that as the game continued, the participants started to have more concern towards how the answer to their question can be accessed through search engines by the other team, thus adjusted their way of speaking accordingly to avoid the expressions most likely to be sought out. The result was that they started to play “trick” by using misleading and ambiguous information in their questions so that their opponents would have no way to obtain the most efficient keywords to filter through the information, or did not even know where to start.

It fascinates me to find through the experiment, how fast people adjust their way of expressing to their condition, with high awareness of the pros and cons of the media they use. The result was that people tended to hide “literally” by giving misleading and even deceptive information, probably because of the limited time they were given to design the questions and it was the fastest technique they can utilize as a reaction of defending.

After documenting the whole process, I watched closely at the footage and highlighted the moments where such behaviors occurred, and analyzed the language used by the participants at those moments. The tactics of “making tricks” fell into several categories, such as omission, misleading, ambiguity, paradox and concepts exchange. The process of analysis found here.

Artificiallly intelligent

So the computers talked!

This is an experiment parallel to the lamp project that plays with the idea of conversational interaction between people and daily objects. Iterations were made to test varies forms of expression a computer can take. In this case I tried to distribute some donuts to the computers and asked about their preferences. The computers were programed as such that the first talks through language, the second graphical forms and the third digitally synthesized human speech.

“Reading in public” continued…

Following up the issue of “reading in public” from last post, I started to think about alternative forms reading devices can take that embrace the spirit of both classic and new media. The first object is what I call a “hybrid book” that synchronizes and displays what you are reading to your surrounding. The mock-up created quickly with an iPhone embedded into the book cover was tested last weekend on the Red and Gold metro lines in Los Angeles. It turns out that the book tends to create more seduction in the Red line that runs underground, while on the Gold line, people are more willing to glue their eyes to a vanishing point outside of the windows. It vaguely indicates that the idea functions better in an enclosed environment with views limited and sceneries scarce.

However, one may question about the practicality of the book and it seems to be too much of an altruistic thing with less regard to the reader’s experience. The next iteration of this, will probably push into a selectable “broadcasting” content instead of an automatic one, combining the note taking of the book reader and the sharing of real time knowledge.

The robotic lamp

Just finished testing a hobby project I’ve been working during the summer, a robotic lamp that responds to human speech. Below is the first prototype(lacking of a shade that can withstand its hilarious act).

Early this summer I went to a workshop by Joseph Tepperman about automatic speech recognition and am using a processing library called VOCE which is based on the speech recognition model Sphinx.

The current interaction between me and the lamp still seems more like a commander- receiver relationship and am not very satisfied with the servantal behavior of the lamp. I’d like to see the lamp blend in the domestic environment with a strange identity: an animated furniture, an observer of human condition and a sincere listener to every conversations happening in the space.

Train passing…

Celebration of the project “Friends of Friends of Friends of Friends” before setting up for the parking experiment: clustering everything we got and taking a family picture of them. Amusingly, a train coated with McDonald ad passing by, superimposing a mood to the scene.

Photo credit: Shona Kitchen