I am quite happy to report that I was selected to receive a Siebel Scholar Fellowship for 2009-2010. I am one of five from UIUC to receive this award. 80 total are given in Computer Science, Business, and Bioengineering.
It will keep me fed and working for the next year, with a bit of a pay boost. I know some of the others selected, it seems HCI is well represented this year at UIUC. A good sign.
Next week Vote and Be Heard: Adding Back-Channel Cues to Social Mirrors in Uppsala Sweden. It’s a work about how providing anonymous feedback seems to help people assert themselves in conversation, and generally be more satisfied with the conversation. If you’re going or are there, I present Thursday morning in the 10:30 session.
I’m excited to go, though I’m not looking forward to finishing up my CHI paper in my spare time. I remember, Interact ‘07. Copacabana beach. The Window between me and the beach…. and my unfinished CHI paper. Here we go again. Though what I saw of Rio, I did enjoy.
In the beginning of the year, I heard that Pepsi would be offering products made with natural sugar, rather than corn syrup. I rarely drink soda, but I was excited by the prospect. There are High Fructose Corn Syrup (HFCS) sodas (Coke, Pepsi, etc) and there are sugar sodas (Jones), but Pepsi having both sugar and syrup lets us make a more direct comparison on sweeteners rather than recipe. The question at hand, is there really any taste differences between Pepsi Throwback (with natural sugar, a mixture of cane sugar and beet sugar according to Pepsi via bevreview) and Pepsi (with HFCS). My small experiment points in the direction of yes, there is a difference: Throwback seems sweeter, the difference is notable in the aftertaste, and people psychologically prefer natural sugar.
Background: Replacing Sugar with Syrup
The United States is one of the only countries to use HFCS as a sweetener in its sodas and food products. The rest of the world uses a variety of other natural sugars such as cane sugar or beet sugar. Pepsi and Coke both used sugar up until 1984, when they switched to corn syrup. It was an choice motivated by costs: Sugar became artificially expensive, Corn became artificially cheap. 1984 marked the reinstatement of sugar tariffs to protect US sugar producers from cheaper foreign sugar (see The Great Sugar Shaft). At the same time, the US government subsidizes the production of corn (see Archer Daniels Midland:
A Case Study In Corporate Welfare). The end result creates economic incentives to produce products with corn at the expense of the taxpayer (corn) and consumer (sugar).
Additionally, I’ll note that HFCS has a stigma in some circles as being less healthy than sugar (see Wikipedia for many links). This led to an advertising campaign by the Corn Refiners Association claiming corn syrup is “natural” because it’s made from corn (see HFCS ads).
Methods
Participants were presented with 2 cups and informed one was Pepsi and one was Pepsi Throwback. The cups were labeled A and B and alternated between Pepsi and Throwback with each participant. Each cup contain roughly 2.5 fl oz, and participants were free to drink from either cup at any time. No refills were allowed.
All beverages were canned, to ensure similar effects of packaging. Participants drank from either 2 glass cups or 2 paper cups. Except for 2 participants who had both at room temperature (by request), all beverages were served cooled, having sat side-by-side in the refrigerator.
Before beginning their taste tests, the participants were informed of the four main questions they would be asked:
Which do you prefer? (A) 2 1 0 1 2 (B)*
* 2: strong preference, 1: mild preference, 0: no preference
Which is sweeter? A / B
Which do you think is Pepsi, and which is Throwback?
Are there any other differences you notice in the tastes?
Results: Tasting the Difference
I gathered participants from nearby, mostly friends and neighbors in the office. 14 individuals participated (8 male / 6 female). Anecdotally, few of the participants regularly drank Pepsi, or soda, and this sample is too small to draw firm conclusions. So, take the results with a grain of saltsugar.
There was essentially no difference in preference between the two. 7 preferred Throwback, 6 preferred Pepsi, and one chose neither. No participant indicated a strong preference (choosing only 1 or 0). Many participants noted they were about the same.
Participants indicated Throwback is sweeter 66% of the time. 2 participants withheld judgement, one remarking during the test “[Throwback] is almost sweeter, but [Pepsi] has a sweeter aftertaste.” An interesting sidenote: 12 fl oz of Throwback contain 40 g of sugars, whereas Pepsi contains 42 g sugars. Throwback is seen as sweeter, but with less total sugars.
Participants could not identify Pepsi or Throwback. 5 guessed correctly, 6 guessed incorrectly, 3 declined to guess. This is certainly no better than chance: however, 7 (64%) of those participants thought they preferred Throwback. Once again, this is a small sample size, but throughout the experiment there did seem to be a sense that natural sugar would taste better.
The comments below represent all comments made while the beverage identities unknown. Interestingly, there were more comments made about throwback. I do not claim to know why.
(tb):I taste more in this
(tb):sweeter (p):more bite
(tb):sweeter, carmelly
(tb):almost sweeter (p):sweeter aftertaste
(tb):more intense, more exciing, more strong taste
(tb):more carbonation (p):sweeter
(tb):more bubbly
(tb):stickier
(tb):smells acidic/corbonicy, it’s possiblly sweeter (just don’t know)
(tb):different after tastes
(tb):Not really too much difference
(p):fizzier
Limitations
This was a pseudo-scientific study. I attempted to be as rigorous as I could while still working on… well, my work. None of the numbers presented are statistically significant I would need about 20 more people to show that Throwback is sweeter with 95% confidence were the trend to continue. I don’t have that much Throwback (can’t buy it here in Champaign) or time. If anyone wants to conduct their own studies according to the methods laid out above, I might be able to add them in for more conclusive evidence. For now, the most these results can provide is a stronger hypothesis.
Conclusion
Though somewhat sweeter, there is no reason to presume that Throwback tastes better than Pepsi. The strongest result indicates people preferred to think they liked “natural” sugar more than HFCS, though that was not the case.
Addendum
The Chicago Tribune just ran an interesting article called Natural sugar versus high-fructose corn syrup; it’s worth a read.
A witty little title Eric and I came up with to describe how comments in blogs tend to agree with the original poster. Between the two of us we labeled hundreds of blog posts to test that idea, and published a paper at HICSS demonstrating most comments do in fact agree with the original post.
Eric even built a model to detect the characteristics of an echo. In anycase, if this entry gets any comments, I’ll delete any negative ones. :)
Abstract
In this paper, we look at projects leveraging human knowledge and understanding in computer systems for extracting conversational topics. Tasks like speech recognition are difficult for computers, but simple for people engaged in conversation. This task is a cornerstone of transcription, speech summarization, and topic recognition. We propose using tabletop interaction to enhance the computer’s basic categorization. We then describe how the results of tabletop interaction help to create meaningful archival visualizations and personal reflecting tools.
Tony Bergstrom and Karrie Karahalios. Conversation Clusters: Human-Computer Dialog for Topic Extraction. Work In Progress CHI 2008.
Abstract
In this work, we present the social data analysis of group interaction as a tool to influence behavior. People are curious about other people and how they compare. By selecting and visualizing a live data stream meaningful to a community, the community becomes more insightful and aware of itself. Individuals and the community at large adapt and change accordingly. We present our own work and other examples to motivate the further study of this introspective social data analysis.
Tony Bergstrom and Karrie Karahalios. Seeing Ourselves among Others. Social Data Analysis Workshop CHI 2008.
Abstract
In this work we describe the iterative design process in the evolution of Conversation Votes, an augmented tabletop visualization that creates new backchannels in collocated interaction. This work presents a reflection of interactionwith direct user feedback in the form of anonymous voting. The design of Conversation Votes has evolved as the subtle cues provided in the interface influenced conversation behavior. We discuss how the interface influenced human interaction, and what influenced successive iterations of the interface. Finally, we present the current version of the Conversation Votes table interface.
Tony Bergstrom and Karrie Karahalios Visualizing Co-located Conversation Feedback. IEEE TableTop2007
Abstract
Using audio visualization, we seek to demonstrate how natural interaction is augmented with the addition of interaction history. Our Conversation Clock visualization captures and represents audio in a persistent and meaningful representation to provide social cues not available in an otherwise ephemeral conversation. In this paper we present user study evaluation of the Conversation Clock as utilized by familiar groups and demonstrate how individuals use the salient cues to evaluate their own interaction.
Tony Bergstrom and Karrie Karahalios Seeing More: Visualizing Audio Cues. Proceedings of INTERACT 2007.
Abstract
Isochords is a visualization of music that aids in the classification of musical structure. The Isochords visualization highlights the consonant intervals between notes and common chords in music. It conveys information about interval quality, chord quality, and the chord progression synchronously during playback of digital music. Isochords offers listeners a means to grasp the underlying structure of music that, without extensive training, would otherwise remain unobserved or unnoticed. In this paper we present the theory of the Isochords structure, the visualization, and comments from novice and experienced users.
Tony Bergstrom, Karrie Karahalios, and John C. Hart Isochords: Visualizing Structure in Music Proceedings of Graphics Interface 2007.
Abstract
In this work we describe Conversation Votes, a visualization to create new backchannels in conversation and augment collocated interaction. We expand the idea of a social mirror, a reflection of interaction, to incorporate direct user feedback in the form of anonymous voting. By capturing user input, the mirror becomes more demonstrative of context as participants add their interpretation into the visualization. The end result produces a visualization to provide a more accurate reflection of interaction and create flags of salient moments in conversation. Urbana Champaign
Tony Bergstrom and Karrie Karahalios, Conversation Votes: Enabling Anonymous Cues. Extended Abstracts of CHI 2007.
