Link to Indiegogo campaign.

So what’s in that Twinkie? Sure, there’s the familiar fat and sugar, but what about the thiamine mononitrate? riboflavin? anything else?  Most people choose to rely on the food labels mandated by the FDA to find out what’s in their food, but the folks at TellSpec claim that they have a better solution.

The Tech

The TellSpec is claimed to be a portable hand-held  spectrometer that can be simply waved over foodstuffs to collect data.  These data are then sent to something in the “cloud” which fires back with relevant information about what’s in the food.  They have provided very few details on how their product is supposed to work, so it’s a little hard to judge how feasible their particular implementation is.  It’s definitely an interesting concept though, so let’s take a look at what spectrometers are and how they work. What could such a device actually do?

The TellSpec supposedly uses Raman Spectrocopy which is an analysis method in which laser light is fired at an substance to excite the molecules in that substance causing them to emit photons of their own.  Depending on what kind of substance you’re dealing with, these re-transmitted photons will have varying amounts of energy which will give them different wavelengths.  By measuring the intensity of the returning wavelengths of light, information can be gathered about the substance.

This is sort of explained in the project video, but a much better (albeit dorkier) explanation can be found here:

Any time I see a magical device that performs a task by simply “waving it around”, my skeptic sense starts tingling. However, with a little more research, I found some evidence that lead me to think that aspects of their proposed device might be possible.  There are a number of companies marketing handheld Raman spectrometers such as the TruScan which is a handheld device meant for analyzing materials in pharmaceutical manufacturing environments.  These devices require that the sample be pressed up against the scanner, but even if TellSpec changed their product to require this I don’t think backers would be majorly upset.

(photo from here)

So yes, it is possible to build a Raman spectrometer that can fit into a hand (though many real-world examples are many times larger), yet TellSpec are really stretching what spectroscopy can do.  One of the biggest challenges with this method of measurement is extracting useful information from the data.  As demonstrated in the video above, every substance you look at will have some kind of “fingerprint”, but if you have multiple substances in a single sample, you will have multiple fingerprints added on top of each other.  There is a very good chance that the sum of two fingerprints will look like a third  entirely unrelated fingerprint.  It’s like someone saying they have 87 cents in their pocket and asking you to guess how many nickels they have.

This is why these scanners are usually used to test for product purity.  If a measurement returns results that are consistent with what you’re expecting, it’s very likely the right substance. Otherwise, there may be contamination.  However, with absolutely zero context to draw from, it becomes impossibly difficult to determine which out of an infinite number of chemicals were combined to make the spectrum that you’re seeing.

Of course, this is assuming that the chemicals are mixed together very well.  Another one of the challenges of Raman spectroscopy is dealing with non-homogenous samples.  For example, if you point your TellSpec at a steak, pointing it at the meat, bone, or fat should yield very different results.  Current equipment doing this type of analysis on food has automated methods for conducting a broad survey of the food being tested.  I found one such study that discussed analyzing lamb meat for pork content using an automated system to get a representative scan of multiple points on the sample.

Both of these problems are purportedly solved by the TellSpec “analysis engine”.  The team claims that the system can leverage data from other TellSpec users to provide more thorough information about the food being scanned.  For example, if you scan the outside of a Twinkie, you’re only going to find information on the golden cake exterior.  However, if the analysis engine can somehow differentiate between the exterior of a Twinkie from other snack cakes, it can use more thorough scans from other users to display what you might expect to find in the cream filled center.  Their CTO explains this here:

…if this same food was scanned earlier by another user, who scanned it more thoroughly or, in this case, cut the chocolate in two, then the user may still be informed about the likely presence of nut allergens.

At least I think that’s how it’s supposed to work.  It all seems very complicated, and they haven’t really discussed how the system is going to know exactly what kind of data is being scanned.  What if the exact same golden cake exterior is used on an entirely different snack?  Their app demo doesn’t show any instances of adding context (“is it Twinkie shaped? [Yes] [No]”).

On top of that, if the system is leveraging some “social” aspect to gather relevant information about the thing you’re about to eat then you’re just moving your trust from a federally mandated food label to some online social network made from scanners operated by the average consumer.  Doesn’t seem super reliable.

While all of this doesn’t add much credibility to their claims, the most implausible part of the project by far is the $250 price tag.  Although I wasn’t able to find a price quote for the TruScan, a quick google search turned up a number of results for similar products that should give some idea such as this spectrometer which retails for the ultra-low price of $28,045.75 used.  These devices aren’t cheap.  This is why they’re presently only being used in professional environments such as laboratories or factories where they will serve a very important purpose protecting thousands of potential victims from material contaminants.

I’m not going to completely rule out the possibility that what they’re proposing is physically possible, but it does seem a little unlikely that a hand full of people on Indiegogo could undercut an entire industry by 99% on price.

Marketing

Just check out their demo video:

The marketing for this product is really low hanging fruit.  I could spend a few paragraphs analyzing why I think their video is fake including the fact their prototype indicates that an apple is a great source of omega-3 fatty acids, a group of fats commonly found in fish and seed oils:

but fortunately, Dr. Stephen Watson, the CTO of TellSpec Inc. has already admitted that the video is a “simulation” in a reddit comment:

They’ve also added a disclaimer to the video clarifying that it isn’t a real demonstration.  That’s a shame because I was really looking forward to eating the label on that package of tortellini

So what about the rest of their page?  They show a handful of computer and form-factor models of the product, but no circuit boards are in sight.  That’s a good thing too, because with all of that plastic structural ribbing in their 3D model, there probably wouldn’t be any room for a circuit board or battery anyway:

Seems like an awful lot of effort for a non-functioning prototype.

Okay, so we have a fake video demonstration of a fake prototype that likely doesn’t even closely approximate what the final product will look like (I mean, it doesn’t even have a battery charging port), but if there’s one thing that really screams SCAM here, it’s their choice of funding options.

Indiegogo allows “flexible funding” which means that the project creators get their money even if they don’t reach their fundraising goal.  This is normally intended for charity type projects where, for example, someone would love to build a new library for the community, but they can probably find use for any amount of money people are willing to part with.  This system works because these types of projects typically don’t offer very valuable perks to their backers (t-shirts, stickers).  Giving away a not-yet-existant $250 food scanner might be a little difficult if you don’t raise the money required to finish developing it.

These guys have selected the flexible funding option which means that if they can’t raise their full $650,000, they still walk home with over $35k.  The Indiegogo rules are very clear on this and state that flexible funding campaigns that do not reach their goals are still required to fulfill their perks.  That seems a little odd.  If the TellSpec folks don’t raise enough money to bring the product to market, they’re still required by the terms of service to deliver at least 164 units to their backers.  So what were they thinking?

Uh…. okay.

Can they do it?

This is where I really struggled in writing this article.  There are plenty of cases where we find snake oil projects like this from people who are either totally deluded or very clearly setting up a scam.  I was ready to pass off this project as one of these until I saw who was behind it.

Stephen Watson holds a Ph.D. in mathematics from the University of Toronto and has held a tenured position at York University in Toronto since 1993.  I spent some time trying to dig up some dirt on this guy, but he checks out.  He has a LinkedIn profile, a university staff page, and even a decent rating on RateMyProfessors.com.  He looks like he’s the real deal.

Likewise, I found Isabel Hoffmann (sometimes Hoffmann-Miles) on LinkedIn, she’s the President of the Iberian Chapter of the World Anti-Aging Academy of Medicine, she’s been interviewed on Portuguese TV, and I found multiple articles online that mention her accomplishments as a business person.  She did her part to fight the Children’s Online Privacy Protection Rule.  She’s even listed in Who’s Who in Canadian Business (2001), though she might have lost that status when the dot com bubble burst:

The point is that both of these people are indeed real people.  They have multiple references to their names linked to their pictures online.  You can doubt their qualifications all you want, but whatever this project is, it isn’t a simple scam.  These are real people who can easily be tracked down should legal proceedings become necessary.

Heck, their credentials are pretty impressive!  Though he doesn’t go into much detail on how the product works, Watson’s background in mathematics is certainly appropriate for someone creating an online data analysis engine. And yes, Hoffmann crashed a dot com, but it had a market value of $75M before it went under, and she apparently listed in Canada’s top 40 young CEOs before that (though I wasn’t able to verify that).  She’s clearly had some experience in business.

So what are these people doing!?  I have a few theories.

What they certainly are not doing is shipping this product in a year for $650,000.  Even with an alpha prototype, the kind of money required to bring something like this to mass production is well over a millions dollars.  With no alpha prototype, they still have to finish up their costly research and development.  Watson has made some mention of successfully trying out their algorithm on “three different models of spectrometer”, but any specific details are being pushed behind the “intellectual property” curtain:

Our approach is excellent at overcoming both noise and fluorescence, while our technique for handling the scattering of ambient light is described in our patent application and, as you note, is needed for reliable performance. I can’t describe much more about our algorithm, or we might just as well tell Samsung how to do the whole thing!

I think there are two possibilities here. Either this pair have tons of business and technical experience, but simply don’t understand the difficulties of bringing a product to market, or they’re already well on their way to drumming up some R&D funding and are just using Indiegogo as a method for gauging interest and generating buzz about their work.  Some approximation of the product they’ve promised is definitely possible, but they haven’t shown a single bit of evidence that their solution actually works. Your decision to back should be based entirely on how much you trust these individuals.

There is one thing I do know for sure though.

Twinkies are delicious.