Ethnography Assignment Phase 1

I found myself in a rush, trying to find a group to study throughout the semesters. I wanted to find a group that LARPed (Live Action Roll Playing) as shown in the comedy movie, Role Models (2008). With the help of a student in the class, they pointed out that Cepheid Variable might be the club I was looking for, and was having a meeting that Tuesday.

I had no idea what to expect. I wrote down the following things that I figured the group would be like.
· I knew it would be full of Nerds.
· Possibly, if not mostly book Nerds, enveloped in fantasy.
· I figured the Male/Female Ratio would be about 10-1.
· Physical appearances that of really skinny, or fat.
· Most would have Poor hygiene, greasy hair.
· I also assumed the club would be full of people quiet and shy.

I then looked them up via: Google. And found their website. On their homepage they explained a lot about themselves. Cepheid Variable is an organization devoted to the support and promotion of all things science fiction, fantasy, horror, science, and technology, and the support of the community that grows around them through an atmosphere of acceptance and kinship. Cepheid seeks to promote creativity, enrich imagination, and build active learning opportunities through a variety of educational programs, demonstrations, and presentations.

With that I had I felt I had a pretty good expectation on how the group and meeting would be.

Day of the Meeting

First off, since it was the second week of club meeting, apparently Rudder Tower hosts about 20 clubs at 8:00 on Tuesdays. The hall way was packed waiting in line for all kinds of clubs being held on the above floors. It turns out that the stairs were closed because the second floor was under construction, causing the pile up of students. It took me about 10 minutes waiting in a line to get to an elevator, just to get to the third floor.

I finally get to the third floor, and found the meeting room. My first expectation of the club being full of nerds was spot on, not a big deal. But I was dead wrong about how many members there were. I counted at least 110 people trying to fit into a little class room that could hold maybe 60. And the noise was at a high level decibel, which completely shattered my assumption that people would be quiet and shy.

Everyone seemed to have there on clicks (groups), in which they hung out with. When the room began to become more full, there were empty seats, and people would blatantly tell people standing around, “Hey, there’s a seat right here, next to me!”; which made my bias opinion that people were shy.

The meeting started, and it turned out that we had to LEAVE the building, because as a group, they would be violating fire codes, and needed to find a new setting. They called it “mass exodus part II”; they apparently did the same thing the previous week. Since the group had to deal with the elevators again, we had to wait another 10 minutes or so waiting for everyone to reach the bottom floor to move outside were the meeting officially started.

During those 10 minutes, I observed something very noticeable. Most of the people would randomly do a set of things, trying to amuse whoever they were talking to.
· Move their arms up and down like, like a walking monkey, or some weird jester to that effect.
· Make a weird noise while telling a story.
· Try to crack a joke, which no one seemed to understand, and didn’t laugh.

Once everyone was in the lobby, we proceeded to leave the building, and some how about half the group started to do a Congo dance line, where you put your arms on a person’s shoulders in front of you, and someone does the same to you behind you.
            

The group found a spot outside where they proceeded with the meeting that had been interrupted due to a large number or participants. From the meeting I found that, there are many groups of people that have little mini clubs. Some with the game of “Magic” and other board-games, some that deal with Video gaming, movie nights at someone’s house, fantasy clubs, ninja clubs, and sci-fi clubs. All information is posted on their facebook website and if anyone wanted to join the club, they only needed to sign up via facebook, and join them at their separate meetings.

After the meeting was over, about half of the group attending the “D-raid” event. Which is mass movement to Double Dave’s Pizza, where everyone gets to eat pizza rolls.

Blog #7 Performance Optimizations of Virtual Keyboards for Stroke-Based Text Entry on a Touch-Based Tabletop

 Performance Optimizations of Virtual Keyboards for Stroke-Based Text Entry on a Touch-Based Tabletop
Authors: Jochen Rick - The Open University, Milton Keynes, United Kingdom
location: UIST’10, October 3–6, 2010, New York, New York, USA

Hypothesis:
The Author wants to find a better way to write on touch pad technology. To better recognize, and use the human hand on tablet like user interfaces. He thinks that shape writing would be better then using the conventional Qwerty keyboards on tablets and cellphones, since there's no places for the hand to rest, and properly type the speed the user is used too.

Methods:
Testing began with 8 Participants s (6 right-handed, 2 left-handed) to determine speed and efficantcy using different strokes. Participants completed a series of stroke sequences through four points, in order from 1 to 4. A four-point sequence was chosen so that each stroke sequence included a beginning (beginning a stroke), a middle (continuing an existing stroke), and an end (completing the stroke by stopping at the end point and lifting the finger). They were evaluated on the tasks to see the differences in the strokes, and accuracy. The Authors then calculated the speed at witch the users would be typing on known existing layouts for cellphone keyboards.

Results:
From calculating all data gaind from the users, he showed a chart in wich using existing layouts for cellphone and tablet keyboard deal with input. It turns out, that using the Qwerty keyboard really isn't as fast as we thought, because others would be faster. (Compare Qwerty to all the others in tapping speed). The model introduced in this paper predicts that a switch from tapping keys on Qwerty to stroking them will bring a speed improvement of 17.%. While that is a large improvement as far as keyboard performance goes, a substantially larger gain can be achieved by simultaneously switching to a different layout. Stroking with OPTI-II is predicted to be over 50% faster than tapping with Qwerty.



Discussion:
The model of stroke timing introduced in this paper provides a useful tool for comparing the performance of different keyboard layouts for stroke-based text entry. And it gives enough evidence that there is other tools in the tech world that may increase the speed of using Cellphone/Tablet keyboards.

I thought it was intresting to know that i could use a different keyboard and type faster... But i really dont want to learn how to type on a new surface. In WWII they changed the keyboard to the QWERTY system for a reason, or slowing the typing down. I knew that qwerty wasn't the best, but i also know that it was men to create accuracy (and not to jam the type writers). Maybe going back to the old system prior to WWII could become advantages to use, or even some of the keyboards listed above in the picture.

Blog #6 TurKit

TurKit: Human Computation Algorithms on Mechanical Turk






Greg Little, Lydia B. Chilton, Max Goldman, Robert C. Miller

Greg Little: PhD student at MIT working in the User Interface Design group.
Lydia Chilton: graduate student at the University of Washington and attended MIT prior to that.
Max Goldman: graduate student at MIT focusing on user interfaces and software development.
Rob Miller: associate professor at MIT and leads the User Interface Design Group there.
This paper was presented at the UIST '10 Proceedings .

Hypothesis:

The Authors have purposed a Mechanical Turk application used for human computation. The Turk application makes a toolkit that provides a framework to build more complex algorithms. The authors believe TurKit will make human computation more effective than it currently is by enabling complex algorithms.

Methods/results:

First, they talking about the TurKit Scripts. The TurKit Script is an extension of JavaScript that introduces functions for interacting with the MTurk platform. Script adds support for the functions waitForHIT, prompt, vote, sort, fork and join for parallelism.

Second, they discussed is crash-and-rerun programming which TurKit Script is based on. Every line that is successfully run is stored in a database. After a crash, the program will automatically re-run from the beginning. The authors needed to adjust the crash-and-rerun script because some of the functions being called actual cost money. They had to develop a once function that tells the script to look for an output to this line before running it to see if it had already been run.

For the third part of the paper, i would say it was the authors explenation of the programs interface. its had a onling web interface. The interface also has facilities for managing projects, editing files, viewing output, and managing the execution trace. And when managing projects, the easiest way to use TurKit, is to use the
prompt function. This function shows a string of text to a turker, and returns their respons.

Finally, The researchers provide examples of where TurKit has been used to great affect. They also showed what cannot TurKit currently do in standard to MTurk. These include iterative writing, text recognition, and decision theory experiments. All of these examples take advantage of the crash-and-rerun nature of TurKit and allow users to build off of other users' answers.

Disscussion:

There is still more experimenting really needed to make there hypothesis valid, but overall, i really do think they are close. I hope that in one day, they do make my programming more effective, and even expand the industry into "true" AI. Overall, i do think it was a success, they were able to many experiments on examples on how the tool helps programmers.

I would like to play with the crash-and-rerun programming (if i knew where to find this) that they used in TurKit. The TurKit crash-and-rerun programming model makes it easy to write simple scripts, but users have uncovered a number of usability issues. They explored a new territory in human computation by supporting iterative improvement allowing users to build off of other users to get maximum feedback. The fact that they have assistance for parallel programming, may prove to be a usable tool for parallelism research.

Blog #5: A Framework for Robust and Flexible Handling of Inputs with Uncertainty

A Framework for Robust and Flexible Handling of Inputs with Uncertainty

Authors: Julia Schwarz, Scott E. Hudson, Jennifer Mankoff, Andrew D. Wilson.

Julia Schwarz: BS in Computer Science from the University of Washington &

working on her PhD in HCI at Carnegie Mellon.

Scott E. Hudson: BS and MS in Computer Science from Arizona State University,

PhD in Computer Science from the University of Colorado, professor at Carnegie Mellon teaching and researching HCI.

Jennifer Mankoff: BA from Oberlin College and a PhD in Computer Science from the GIT (Georgia Institute of Technology), professor at Carnegie Mellon teaching and researching HCI.

Andrew D. Wilson: BA from Cornell University, MS and PhD from MIT, 

Senior Researcher Microsoft Research

Hypothesis: 

The authors developed a method  to allow a more interactive environment for humans to interact with computers by handling ambiguous user inputs. The current system of handling interactive user input not fill the requirement of the users, and reducing the interaction with the computer and the capabilities of the computer. The Authors believe by handling user inputs they can enabled their system to take in and manage with all sorts of uncertain inputs thus making a computer more us-full then before those with motor impairments.

Methods:

First the Authors needed to compare conventional and uncertain input from the user. The conventional input handling framework provided four major capabilities:

• (1) modeling of inputs, by providing a way to record all the relevant details of what input happened 
• (2) a process for  dispatch of those events – deciding which interactor object(s) should receive and handle a given input 
• (3) interpretation of those events by the interactors
• (4) action by the interactor

With the framework created, the user was given tasks in dialog boxes, scrolling, and text boxes that take an audio input and transcribe it to text. 

Results: 

While testing the framework themselves the found it operated as expected. It was able to understand and handled uncertainty very well. Giving the user a greater interaction between humans and computers. Users with motor impairments saw large a reduction in errors when using their accurate probabilistic tracking method of handling inputs.

The others purpose of the research came to be a success. The challenged the current attitude towards the handling of interactive user inputs and demonstrated the power of a more probabilistic method. 

Discussion:

To design a system that may actually understand whether or not a user had meant to push a button or not is highly advanced. I'm a little impressed that they could design a system to recognize that little detail. I would like to see in the future some more detail about how the computer interacts with the human. Cars even do it today by watching the iris of the driver and sweat on the figures tips to calculate stress. It may be a little to "Big Brother" as we make programs more aware to what humans mean do when they perform a action, but i find it an advantage on the usefulness of technology.

Blog #4 Gestalt: Integrated Support for Implementation and Analysis in Machine Learning

Gestalt: Integrated Support for   Implementation and Analysis in Machine Learning

Authors:

Kayur Patel , Naomi Bancroft , Steven M. Drucker , James Fogarty , Andrew J. Ko , James A. Landay       

Kayur Patel: Computer Science & Engineering          

DUB Group, University of Washington                      

Steven M. Drucker: Microsoft Researcher,
PhD in Arts and Media Technology both from MIT.

Naomi Bancroft: BS Computer Science from             

University of Washington and works for Google.                                                                                                                                                   

James Fogarty: BS in Computer Science Virginia Tech,
PhD in HCI from Carnegie Mellon,& Professor @ University of Washington.

Andrew J. Ko: BS in Computer Science and 

Psychology, Oregon State and a PhD in HCI from Carnegie Mellon.                                               

Professor @ University of Washington.                  

James A. Landay has a BS in EECS from UC Berkeley    
MS and PhD in Computer Science ,Carnegie Mellon.
Professor @ University of Washington.

Hypothesis: 

The Authors hypothesized that users would be able to do a better job of writing and debugging code if they didn't need to start with a blank slate. To them that less of a "daunting" task to use their Gestalt Development Environment compared to more traditional methods.  "Gestalt supports two high-level tasks in applying machine learning:  implementing a classification pipeline and analyzing data as it moves through that pipeline." 

Methods: 

They selected a group of 8 participents that they felt suited for the experiment. The participants then created the baseline condition, which was a general-purpose development environment in where they created, edited, and executed scripts. Both baseline and Gestalt used the same data table structure however the baseline data table structure did not keep track of information generated across the pipeline and the participants had to write code to connect raw data, attribute values, and classification results or to create side-by-side visualizations.

The participents were asked to solve two problems.  One queston was about a sentiment analysis and the other was about gesture recognition. Each contained data and five scripts: 

• parsing
• attributes
• splitting
• training
• testing

Then the Authors created working solutions for both Gestalt and the baseline; and injected five bugs to replicate common programming errors into each solution. 

Result: 

As a result, the users were able to find the answer to the bugs unanimously with the Gestalt Development Environment. And found that Gestalt was simpler and easier to fix the common errors. And validated the Authors point that Gestalt will allow users to have a "less daunting" task while debugging. 

Discussion:

As a novice programmer myself, i love using debugging tools that would catch errors before I compile. It makes so much easier, so having something like Gestalt would  not only increase proficiency while programming, it also save loads of time. I'm glad the authors did explain what the limitations of Gestalt were at the end of the article also, it seemed relevant and protected themselves from scrutiny. Having more tools like Gestalt to program with, the better I'd say. 

Blog #3: Pen + Touch = New Tools

Pen + Touch = New Tools

Authors:

Ken Hinckley, a researcher for Microsoft.
Koji Yatani, a graduate student at the University of Toronto.
Michel Pahud, a senior researcher for Microsoft.
Nicole Coddington, a senior interaction designer at HTC who was previously at Microsoft.
Jenny Rodenhouse, works for Microsoft as an Experience Design for Xbox system.
Andy Wilson, a senior researcher at Microsoft with focus on Human Computer Interaction.
Hrvoje Benko, a researcher for Microsoft and focusing on Adaptive Systems and Interaction.
Bill Buxton, a Principal Researcher at Microsoft in Toronto.

Hypothesis:

The Microsoft researchers wanted to see how the user would react to writing with a pen but still interact by touch on the screen. And they hypothesized that with the use of pens and touch would allow the use of new tools and features. Where the pen annotates, and the touch manipulates. 

Methods: 

First, they observed how people would use notebooks and paper. They asked each participant to illustrate their ideas for a hypothetical short film by pasting and annotating clippings in a paper notebook. And then observed them as they filled up the notebook. 

Second they moved to the Microsoft Surface for the physical interface. they explained all the features that one could us on Microsoft Surface. For example, Manipulating, Zooming, Selecting Objects, X-acto Knife (to cut pictures to the users liking), stapler (grouping items into a stack), and ruler.

Results:
The researchers found that everyone had 10 specific gestures as well as design attributes while creating there notebook to illustrate there idea for a hypothetical short film.

•  Specific Roles
• Tuck the Pen
• Hold Clippings
• Hold while Writing
• Framing
• Scraps
• Extended Workspace
• Piling
• Drawing along Edges
• Hold Page while Flipping

As a result to that, they were also able to do the same thing with accuracy using the Microsoft Surface. 

Discussion:

I think that, any time the user can use something as close to physical objects on a computer screen to do a project or fun activity makes it increasingly easier to any user. So the idea to be able to manipulate physical like objects on the surface is a really cool idea. 

Marketing wise, I don't know how expensive an item like the Microsoft Surface is, so i could only imagine one would need a hefty wallet to purchase one. As a learning tool, i can see it being used because of the pen and touch options. Even though the authors claimed that there experiment went to there liking, they didn't show enough info. Maybe they didn't because the facts weren't something measurable, and it was all just observational evidence. 

Blog #2 Hands-On Math

Hands-On Math: A page-based multi-touch and pen desktop for technical work an problem solving

Written by: Rober Zeleznik, Andrew Bragdon, Ferdi Adeputra, and Hsu-Sheng Ko of Brown University , Providence, RI.

Hypothesis:
The idea of solving problems with a pencil and paper versus use of Computer Algebra Asystems (CAS) have many strengths and weaknesses. The Authors main hypothesis is that using a pen and paper computer program can carry over the benefits of using just pen and paper and learn more effectively.  They call this program, Hands-On Math.

Methods:
The first they discussed was the Page Management. The user would be able to work with either a whiteboard which has no spatial management, but it quickly fills up. Or pages like a notebook which has spatial organization. Along with deleting and creating pages, there's a Panning Bar to allow the user to sift along all created pages for easy access. The Folding feature was also implemented to hide or re-open text and mathematical equations without taking up the entire screen (basically like looking at quick notes).

Second was gestures. Under-the-Rock menus which are menus that initially aren't visible till the user needed to perform an operational figure like"∑" to do a sum. Touch Activated Pen Gestures allow the user to use the light pen along with there hands for writing. Since gestures are efficient for executing many types of commands, they required more effort then pressing a buttons, so in those situations they has the PalmPrints options for effectively pressing a button and choosing a design option you wanted to use on the table. 

Third was the math interactions for writing mathematical expressions and creating graphs. The user is able to select individual symbols, then use a gesture to perform a complex operation; without making the computer solving the problem right away and the user missing any information they may want to know/learn about. 

Results:

Participants were overall positive about the system, and noted that it would be more widely used if it were available in a potable form factor like a Tablet PC. Participants had little to no difficult manipulating and writing on the pages, but didn't know how to create or delete the pages. The participants provided mixed reviews about gestures that were needed to preform some actions, and said it felt unnatural to use two hands to preform a action. 

Most participants reported that manipulation of the mathematical expressions were simple enough to be useful to them while doing math-related work. And the idea of working step by step and not just being given the answer is a plus for those learning. But they also wanted higher level of operations than simple algebraic transformations, granted these were university students who were testing the program, and I'm sure they were already familiar with the algebraic math. 

Discussion:

I really like this Hands-On Math, even after just reading about it. I could only imagine that I would have gotten better grades in calculus if i had a tool described like this.  I think it would be ideal even a work environment if they implement a "solve to end button". 

The gesture movements may have been a little complicated, but if the program is able to record movements to your liking as a user; i think that would be a plus for Hand-On Math. Making the system customizable creates endless macros for the program. 

I would see myself buying this if they have it in a portable fashion like many of the participants suggested to make it more marketable.