What Quentin Tarantino can teach us about UX in NGS data analysis

Much of genomics work is a collaboration between cell biologists and bioinformaticians. Given two groups of users with such different backgrounds, skill sets, and expectations, how do you build a tool that satisfies both?

There’s a great scene in Pulp Fiction. Vince Vega and Mia Wallace are sitting in a restaurant. Vince begins to tell a story he heard about her, “Promise you won’t be offended.” Mia responds,

No, no, no. You can’t promise something like that. I have no idea what you’re gonna ask me. So you can go ahead and ask me what you’re going to ask me, and my natural response could be to get offended! Then, through no fault of my own, I would have broken my promise.

It’s worth watching in its entirety. It reminds me of  Fredheim‘s 2011 article in Smashing Magazine in which he argues that one cannot “design” the user experience:

We can design the product or service, and we can have a certain kind of user experience in mind when we design it. However, there is no guarantee that our product will be appreciated the way we want it to be… We can shape neither our users’ expectations nor the situation in which they use what we have designed (emphasis mine).

Hazzenzahl’s Model of UX (Originally published in Smashing Magazine)

He bases this on the model of UX outlined by Marc Hassenzahl. Hassenzahl identifies four general ways in which the attributes of a product can influence the user experience: manipulation, identification, stimulation and evocation. These in turn are grouped into two higher categories: pragmatic attributes and hedonic attributes. Pragmatic attributes are those that relate directly to the functionality of the product. Included in this group are things like the feature set and usability metrics. Manipulation fits in here. The hedonic attributes – identification, stimulation, and evocation – are more fuzzy. These have less to do with core functionality and more with the emotional response of the user. To what degree does the product surprise the user with some new, unexpected functionality? Can the user express himself through the product? How does it connect with past experience?

UX cannot be designed because UX depends not only on the product itself, but on the user and the situation in which they use the product.

The designer has some element of control over things, less over others. Manipulation is fairly straightforward – either the product provides the functionality the user needs to carry out some task or it doesn’t. But what about context? Or history? Fredheim points out that when a user encounters a product for the first time, she first compares it to past experiences. “Mint.com, eh? I sure hope it’s better than my current bank’s website…” The designer can anticipate some things, sure, but he can’t control how past experience has worked to shape this new user’s expectations. Nor can he control the user’s emotional state. Is it the start of the work day or the end? If it’s the morning, has she had coffee yet? How distracting is the environment she works in? All of these things – all external to the product itself – will work together to affect how the user reacts. What, then, does it mean to design for the user experience (as Fredheim says we should be doing)? There isn’t a clear answer. The closest he says in conclusion is that we should design with Hassenzahl’s four attributes in mind:

It has been suggested, for instance, that UX is the sum of certain factors, such as fun, emotion, usability, motivation, co-experience, user involvement and user engagement… In turn, we must address some of these factors when we design for UX, depending on how we want our product to be perceived. If we want an application to be fun, then we need to add some features that will entertain; a joke, a challenging quiz, a funny video, a competitive aspect or something else. We should keep in mind, however, that, as designers, we can never really predict that the application will be perceived as fun by the user. Users have different standards, and sometimes they aren’t even willing to be entertained.

In short, designing for user experience means knowing your users and managing your expectations about the results. Not a particularly satisfying answer, is it?

Here’s my take on it:

First some background. SBG has a product called IGOR. IGOR is a cloud-based platform that gives researchers the ability to run computationally-intensive analysis of DNA from their laptop or tablet.

One of our big UX challenges comes from the fact that our two main user groups – cell biologists and bioinformaticians – come from markedly different backgrounds with different expectations for what they want out of IGOR. Both groups want to find points of significance within incredibly large datasets, but each has found a very different way of going about it. Much of a bioinformatician’s work is done at the keyboard, working from the command line and writing custom scripts to send the data through open-source apps like TopHat and Mosaik and FreeBayes. The work of a cell biologist, on the other hand, ranges from preparing samples at the bench, to analyzing data, and interpreting the results, most often using commercial tools that provide a richer interface layer.*

It is unrealistic to expect a cell biologist to have the computational experience of a bioinformatician. It is also unreasonable to ask a bioinformatician to be content with a wizard-based UI tailored to biologists. The competing needs and expectations of the two groups support Fredheim’s point: A “designed” UX that produces a positive reaction will likely produce a negative reaction in the other.

So what’s the solution? How do you plan for both types of users?

I like to think that I and the crew of mad geniuses I work with have come up with a good response to the weaknesses of Fredheim’s thesis: we design for experience in a way that lets the user create the type of experience she or he wants to have.

Tomorrow I’ll talk about the ways we make that happen. I’ll do it by looking at three things: the ways in which collaboration occurs in labs, Neil Simon’s play The Odd Couple, and an awesome little book of hand-drawn maps called Mapping Manhattan.

*There are exceptions to both, and I’m sure the community will give me shit for oversimplifying things :).