Tag Archives: #htk01

POMI2020 retreat from a non-techsavvy perspective

by

Programmable Open Mobile Internet (POMI2020) is an interdisciplinary research program at Stanford University. It is part of a larger initiative called CleanSlate and funded by the NSF. The key players in POMI2020 are Stanford professors largely from the computer sciences, such as Guru Parulkar and from other auxiliary departments, such as the School of Education.

In today’s POMI retreat meeting in Palo Alto, CA the presentations were promising more interconnected and highly decentralized technological frameworks by 2020. Interestingly, however, the issues that advisors such as Bob Iannucci (formerly Nokia), Rick Rashid (founder of Microsoft Research) and Bill Raduchel (formerly AOL) brought up were fundamentally human-related questions, not so techsavvy ones.

In the first set of presentations, Monica Lam (Stanford Computer Science Dept.) and her team of graduate students presented an open-access database platform (SocialLite) that is being developed in order to collect and formalize social and personal information such as friendships, emails, and photos. This is an effort to overcome the limitations in FaceBook-type social networks that only provide “shallow friendships on a centralized datapark and does all this for profit”. Monica’s suggestion is to build more ad-hoc, mobile, multi-player apps that sit on decentralized personal cloud butlers (as opposed to a single public cloud) and communicates with each other via formalized data queries (e.g., FriedOfAFriend-queries; FOAF). The advisors, however, raised the issue that this effort comes down to the ages-old question of how to standardize representations digitally? Or as Bob Ianucci asked which one wins, the algorithm or the human eye? Until the researchers focus merely on computer-to-computer and not as much on computer-to-human communications, there is little hope for a revolution.

Our group from the School of Education, led by Paul Kim (CTO at Stanford) led the second set of presentations. The starting point was an analysis of the mobile industy (based on the evolution model by Fine, 1evolution of mobile market998). A recent survey of industry leaders and developers suggest that the currently expensive and integral mobile technologies will move into a cheaper and more modular market phase, inviting more mobile users (well exceeding the currently already 4 billion!) and trigger even greater interest in content development communities. This cheap and modular market will make mobile technologies increasingly available, especially in underdeveloped countries of the World, like Africa or parts of India. A project paper about this model-based prediction is on its way to publication soon.

Current in-progress educational applications were presented accordingly by Aiditi Goyal and Theresa Johnson (two grad students in our PomiED group): WeClick – an open-source, platform independent clicker app to be used in classrooms; Environmental Sensing Network (ESN) is a sensor-based science project utilizing design-based teaching methods. Questions from the audience were focusing on the costs and wireless communication issues. I was glad -but slightly disappointed at the same time- that the usability and educational aspects (two that I had some impact) did not gain any criticism.

After the first coffee break, Gabriel Takacs’ (Stanford) from the lab of Bernd Girod (Stanford Electrical Engineering; on the picture) presentation was about an automated image tracking & recognition algorithm that performed locally on a mobile phone. His algorithm is quicker and less data heavy than existing solutions which usually send full images over the network. Graduate students of the team demonstrated mobile applications that recognized foreign DVDs or labeled parts of a building. After identification, the system also linked the information with how much others in the person’s social network liked it. Neat! Check out their cool YouTube demo here. My association about this project was that they are doing what the human eye does — preprocess the information before it arrives to the visual cortex and makes it quick and easy for higher level cognition to match features.

The final presentation block of the morning session was by John Mitchel (Stanford Computer Science). He talked about http://courseware.stanford.edu, a work-in-progress collaborative and decentralized CMS. The idea is to work between institutions (e.g., Stanford & USD) and have open-access materials. What I’m not sure was how different this approach is to other systems, say Moodle. Bill Raduchel raised another issue about copyrights, which could be a real issue for open-access. Eric Conner presented a Mobile Courseware application that is an iPhone equivalent of our team’s weClick approach. Three major features: vote, Q&A and course-notes. I’m really looking forward to see this in action and potentially collaborate with John & Eric!

I wish I could have stayed for the afternoon session, as I really enjoyed interacting with these amazing professors and grad students, but I had to get back to San Francisco before the Sun went down.

Life-logging

by | 1 Comment

Scientific studies on memory are usually focusing on how people remember things. Applied technologies are using these models to increase how much or how well we can recall about the information that surrounds us. As both theory and technology develop at an increasing speed, we see how the human mind becomes extended and embodied into its environment (Clark, 2008).

My PhD supervisor, Itiel Dror and Steven Harnad (2008) explained this process as a natural continuation of how all cognizing agents -let them be biological or artificial- are offloading their cognition into different cognitive technologies. The boundaries between the memories in my head and the ones on this blog have become fuzzy. Please, don’t think of me as a cyborg with microchips on my temporal lobe, but rather someone, who searches unfamiliar things in Wikipedia and keeps a good amount of his data in the Cloud.

Total Recall by Gordon Bell & Jim GemmellA new book with the cheesy title of Total Recall, written by Gordon Bell and Jim Gemmell, two senior researchers at Microsoft, show how far this offloading can push the limits of our imagination. They present their MyLifeBits project that digitized all documents, photos, external memories of Bell so that he could truly become paperless and uncluttered ‘both in his head and on his desk’. But they went further and since 2001, Bell has also attached a SenseCam and a GPS on his body to record and log all life events, meetings, trips, emails and telephone conversations that he faces. This is his personal life’s chronicle, which he calls ‘life-logging’. Here’s a long interview video with the authors.

As I’m reading through the book, arguments are firing to convince me that being able to life-log gigabytes of information per month about my own life and smart-search it back in its exactness all that happened with me via time-place concept tags are the best things the modern human can do. To overcome our mind’s limited memory capacity and extend it into a perfect e-memory is the revolution on our doorstep. It is the deliberation of humankind from its mortal biological chains.

Why is it that it somehow makes me feel uneasy? Because it’s just new as computers or televisions were in their times? Maybe. But the first thing that I could think of was a very human psychological mechanism: forgetting. I feel blessed that I can forget and that every day I do so. Even if the research on repression and trauma is still controversial, I feel reassured that most parts of my life will disappear from myself and from other human beings. Not because I’m not happy with my life, but because in most times, I want to reconstruct and not review my own memories.

Memory reconstruction gives me integrity and a sense of who I am. I think the source of my uneasiness about Total Recall comes from a lack of trust in the SenseCam, in contrast to a self-deceitful comfort that my limited mind’s distorted camera has provided to me in the past almost thirty years.

Lastly, the anecdotal evidence which suggests that those who survived a near-death moment experienced a video of their life flickering in front of their eyes makes me suspicious that our mind is in fact not as much limited in its capacity as we thought. It may actually record everything that we encounter, but perhaps what makes us human is that we’re not capable (yet) to do a Total Recall. The real question is still : why?