Tuesday, August 3, 2021

Jason Cohen on the Penn State tea club, Gastrograph, and AI food design


We next met Jason Cohen related to that online meetup theme, a very interesting contact.  He founded a Penn State tea club at my alma mater (one of them), or really an "institute" since it was both a social club and did research.  We didn't talk much about that PSU experience, more about Gastrograph, a food design company, but I'll summarize a little about that part here first.  Related to referencing his background, his personal blog provides a short introduction, with posts filling in other parts, and a book project site on Chinese tea ceremony is here.

That Penn State tea club is interesting to a lot of tea enthusiasts in relation to representing one high potential channel for developing tea awareness.  The form of that club was unique too, really divided into separate parts covering interest in a range themes.  They supported basic tea awareness, of course, and development of effective brewing technique, but also focused on the ceremonial side, and on tea history, with particular emphasis on Chinese, Korean, Japanese, and Taiwanese teas and brewing practices.  They had an extensive collection of teaware for club members and visitors to use, and had renowned tea specialists visit for events and direct support of tea experience.  

All of that was long after my time at PSU, but it would've been great to have it there as a resource when I had been attending.  Jason clarified that his involvement with the Tea Institute ended in 2018, and it later became inactive.  We really didn't discuss how that had worked out, but any university groups cycling through membership and leadership would pose a challenge for continuity.

I haven't been back there in 4 years, and then just passing through

Related to Jason's current career, Gastrograph is a food development research company, now pursuing an alternative direction to standard flavor and aspect preference testing practices.  Jason explained that conventional form:  a research company would set up a testing program, bringing in a lot of test subjects, along with developing prototype versions of the new foods forms they want to test, and would run a large set of tests, carefully collecting as complete data as possible.  

Gastrograph does something that overlaps a bit, but it's different.  They use an Artificial Intelligence based approach, using a bank of prior test data that develops how food and flavor preferences vary in specific areas, and draw on that to project how changes to a current or standard product offering might be received.  Of course there is a margin of error involved, but then some degree of testing could supplement this data driven versus direct testing approach.  And they wouldn't need to get the same preference pattern identified in testing completely right, depending on the context; providing reliable information and direction could be enough.

It's hard to completely get your mind around how this could be applied in practice, and short discussion of that only went so far.  Of course we asked Jason how this might apply to tea, or how they had applied it to tea, and his answer was interesting for placing both their research service scope and loose tea experience.  They've only used the approach in support of RTD (ready to drink / bottled) tea development, and it wouldn't be a great fit for higher quality loose tea.  Jason explained that one reason for this is that variability in the finished experienced beverage version is problematic, that when different people brew the exact same tea variance in water used, brewing process, device, timing, and so on would change their experienced results.  It's helpful, at a minimum, to understand and define what the final food / flavor experience will be, in order to make predictions about locally oriented patterns in preference. 

I suppose one might wonder if development of tea blends couldn't fill in as a middle-ground exception.  Discussion didn't go there, and to some extent what Gastrograph is doing is demand driven, and oriented towards mass produced food items (new versions of potato chips or such).  Tea and herb blends could be a conventional product but not really a mass-market, high demand volume category.

From there we did drift through lots of other interesting examples and details, but for a summary version of what their work is about that's enough.  One might expect that in order to accurately predict local preferences they might need to collect an unusually vast data set to draw on, not just by country but also by regions within different countries, and Jason addressed that it is a main challenge they face.  As one example he mentioned that Germany breaks up into a number of different food-preference regions.  In another case he relayed that they've been slow to develop India as a services base because local food preferences vary so broadly, and by very limited areas, not necessarily by large and uniform sections of the country.  In Italy too, related to that concern; I suppose that's not a surprise.

Years ago when I first talked to Jason I downloaded a Gastrograph app version that supported making spiderweb graphs of food flavor aspects evaluation.  Jason explained that this tool evolved to be the main way they record data in their collection phase, but that it hadn't been developed as a tool for personal use, for example for people to record their own tea reviews or notes.  It's on their agenda, expanding the function and form for that, but the natural evolution of their business direction didn't raise it as an early priority.  Their business isn't focused on tea themes at all, really, or else that would've already occurred.

this would probably be from an earlier version of the app than is presently available

It seems natural enough why specialty tea isn't a suitable subject for this type of preference testing and product development research.  Better loose tea is what it already is; only flavorings and blends actually vary natural product scope.  And it's not a large scale production item in Western markets.  Bottled / ready to drink teas are, and they've done some work related to that.

From there discussing tea culture and history themes was really interesting.  Jason lives in NYC now, and had attended school in PA, but his personal history covered living in a number of other tea related countries, in China and India, and related to spending time in Korea, Taiwan, and Hong Kong.  It was fascinating hearing a perspective on related parts, about flavored tea preference versus better plain tea, a little on pu'er aging issues, and so on.  More of that would have filled in if we hadn't focused as much on hearing about the Gastrograph direction and methodology, and more about the PSU club, but that was a really interesting and completely unique topic, so it was worth that trade-off focusing on it.

Early on I joined from in the car, with my image lighter or darker by streetlight

Always nice seeing Huyen's family too

No comments:

Post a Comment