Poor Little Bill
Regulation is getting a pretty bad rap these days. At the end of January, the Trump Administration announced that for every new federal regulation, two existing regulations should be eliminated (see here and here). Putting aside the immediate questions of logic and logistics of this order, the implication is that regulations are to be avoided or at least minimized. I think of that poor little bill from the Schoolhouse Rock video, sitting on Capitol Hill, waiting to become a law. Now he’s really screwed.
And admittedly “regulation” can often signal oppressive rules, red tape, and excessive bureaucracy — things we want to avoid. But what about positive connotations of regulation? Your body regulates things such as temperature, blood flow, and breathing, and that is ongoing regulation we should all appreciate. Moving from bodies to bodies of government, I’m also glad things such as drinking water and restaurants are regulated, as those regulations make me feel safer when I turn on the tap or go out to eat. We were all mostly pretty happy about bank regulations imposed following the global financial crisis of 2007-2008. So why the love/hate relationship with regulation?
Red Tape (Wikimedia Commons)
The Parable of the Pregnancy Test
Lately I’ve been thinking about regulation in the context of genetic testing, in particular direct-to-consumer (DTC) genetics. I have been asking others what they think about it too: whether and how DTC genetic testing should be regulated and also how third parties, providing interpretation of genetic data independent of DTC companies, fit into this regulatory landscape. One person I talked to gave me a very telling example of why he didn’t think DTC testing should be so strictly regulated. His argument went something like this:
Think of how you can go into any drugstore, buy a pregnancy test, go home and use it, and interpret the results yourself – without any doctor involved. That’s a huge piece of information to seek out on your own! Now with genetic information, why should it be any different?
I’m paraphrasing, but the key point he was making is that pregnancy tests are accessible to the public, why shouldn’t genetic information be? Next I gently pointed out that well, in fact, over the counter pregnancy tests are indeed regulated, by the FDA. They’re regulated AND available. Genetics could be the same way.
What I’m thinking is that when we first started talking about regulation, his mind went towards restriction — i.e., if access to genetic information is regulated it means people just can’t get it. But au contraire: regulation could mean that people can directly access their genetic information, just that there is some regulatory oversight to that process. It’s an understandable confusion, because when the FDA “cracked down” on 23andMe for giving health information, 23andMe’s compliant response was to stop giving health reports. The FDA stepped in; access was restricted. But that’s not the inevitable outcome of regulation. If things had happened in a different order, 23andMe may have only offered health information after seeking and obtaining FDA approval. The rollout would have undoubtedly been slower, but there wouldn’t have been the same feeling of having something yanked away from us due to regulators.
So let’s remember in this climate of hostility towards regulation — it is not always the slap on the wrist, the removal of personal freedom and access. Regulations, when properly structured and carried out, can actually do good and lead to safer and better outcomes. There are many debates to have on to what extent, and through what mechanisms, access to personal genetic information should be regulated. But confusing regulation with complete removal of access is unproductive.
There’s a new genetic sequencing company that wants you to get paid for donating your DNA data to science. Launched in December, Genos’s service costs around $500 and gives customers the ~3% of their genome sequence that codes for proteins, the “exome.” Additionally, Genos’s platform allows users to share their exome data with (academic and commercial) researchers…and get paid for it.
Let’s evaluate what is and isn’t novel about this proposition. First, people have been able to order direct-to-consumer genetic tests for over a decade. Exome sequencing is a little newer and shinier compared to the smaller scale microarray technology used by major players such as 23andMe and Ancestry DNA, but the general idea is well-precedented. Second, it’s not unusual for people to get paid for participating in research, genetic or otherwise. You take time to fill out a survey or participate in an interview, so why not get a token of appreciation (maybe $20 or $50) for your time and efforts. Note researchers and regulators do want these “incentives” to stop short of being coercive, meaning that it shouldn’t be so much money that a person who otherwise wouldn’t want to participate feels compelled to because they want/need the money.
Ok, so DTC data: not new; getting money for research participation: not new. What does seem to be raising some eyebrows (and piquing some purse strings) is, I think, the directly transactional nature of a customer sharing their — already generated — genetic data with researchers and getting paid, potentially much more than a typical participation incentive. Perhaps over and over again, each time they give Genos permission to share their slice of the company database with another researcher.
It seems a bit….smarmy. Over commercialized and over commodifying genetic data at the boundary of commerce and research — granted that research can be academic or commercial. Feeling conflicted about these developments myself, I was glad to see a recent Commentary on the topic in Nature Biotechnology. Written by legal experts, the Commentary reviews the strengths and weaknesses of a system where research participants are paid directly for their data. I’ll summarize the arguments here, in part because the full article is behind a paywall, and also because it allows me to weigh in on parts of the argument.
1. Respect for persons. Research can be lucrative so the people contributing data should be more valued, i.e. via direct compensation.
2. Uniqueness of information. Paying for genetic data recognizes its high value and importance to people.
My note: This is a vote for genetic exceptionalism – the idea that genetic data is somehow special and sacrosanct compared to other types of personal data. To this point I would counter maybe we shouldn’t be encouraging people to attach so much self worth and interest to their DNA sequence.
3. Promotes fairness and equality. Researchers profit and benefit from people’s data, so people should similarly profit and benefit.
My note: Valid point, but this goes against legal precedents in the US that people do not hold a financial stake in their biological specimens, and extracted data, once donated to research. But law and ethics are not the same, so there is definitely something here.
4. Greater good. The Genos model may encourage more people to participate in research, which benefits all.
1. Might decrease willingness to share. Studies have shown that when people would usually contribute out of altruism, offering them something in return actually decreases participation.
2. Undermines individual autonomy. Offering money may coerce people to participate, actually reducing their personal freedom.
My note: see earlier mention of researchers not wanting to offer so much money to would-be participants, for this every reason. Maybe a matter of “how much”, not “if”, when it comes to payout.
3. The problem of valuation. It may be difficult to assign value to an individual’s genome and thus what people are compensated may not reflect the true value.
My note: right, even giving someone $200 for their genome (the upper limit of the compensation range noted by Genos) could fall short of how much that individual’s genome actually benefits the researcher or company.
While the paper does present both sides, it comes off as saying “this is basically inevitable, so let’s strive for the best outcome possible.”
Genos is making people’s exome sequence available to them, as a personal resource, and also one they can “shop around” to different researchers. Part of this is line with an argument I made in Nature last fall, that genetics researchers should offer to give data back to participants if the participants want it. The reasoning being that more people might turn away from traditional research in favor of consumer genomics, which does make the “raw” data available. I wrote about good things that happen when people become “stewards of their genetic data.” With Genos it goes one step further, in that people become their own data brokers.
It begs the question: what is the right way to reciprocate and engage with research participants? The most traditional approach is to rely on the individuals’ altruism to improve the greater good, promising them no individual benefit in return. I don’t think this is the right model anymore, especially when current information infrastructures make it easier to give something back. But is Genos the right model? What about research models that offer people back some bits of interpretation in exchange for donating their genetic data? Sites such as DNA.land or openSNP, for example. Is that an even and fair exchange? I do wonder, especially in our current state of knowledge where genetic data is arguably more useful in the aggregate, to answer research questions using thousands of people’s data at once, versus an individual with their own genome trying to squeeze some drops of meaning from wringing out that double helix. I think such tools might still be leaning more heavily on people’s interests in general altruism than personal gain.
It will be interesting to see how the Genos roll-out goes over the next couple of months. Perhaps they are bringing us closer to Professor George Church’s vision, quoted on the company home page: “It should be our birthright that everybody on the planet should have access to their own genome.” (Oh, but p.s. New Yorkers – it’s not your birth right yet; Genos can’t ship a test kit to you…)
For my dissertation, I have been interviewing developers of third-party interpretation tools for consumer genomic data. These are tools such as Promethease, openSNP, and DNA.land, among many others, where people who have their genetic data file from consumer testing can seek further analysis and/or contribute their genome to research. Even though I’m only a few interviews in, I’m already seeing some interesting themes. This isn’t a formal analysis here, just some of my initial impressions — “field notes,” you might call it.
- Theme 1. Tools are heterogeneous in terms of their creators/developers, functions, purposes, and processes.
Just like with most studies, I have inclusion criteria to identify eligible tools. (Also called “scoping,” this is an essential process for any successful…and completable…graduate student project! We just can’t get enough of the scoping…). My three criteria for tool eligibility are:
- Must allow users to upload (or analyze locally) their genetic data file
- Must return some type of information or interpretation to the user
- Must be active at the time of my study (and there were some tools that failed on this – i.e., they are now defunct)
This is a very user or customer-facing set of inclusion criteria, which I did on purpose. That is, I thought of a user sitting at their computer, having just freshly downloaded their genetic data from 23andMe or AncestryDNA. What tools might they search for or stumble upon as they pursued further self-directed analysis of their genetics? Those were the tools I wanted to study.
At the outset, I was aware I might be grouping together a lot of very different tools, and indeed after months of closer study this is definitely the case. Some tools might not even want to be called “interpretation tools,” as they are more focused on users contributing data to research and just happen to give some tidbits of info in return. I also wasn’t expecting to include some DTC companies, but turns out there are some that also offer analysis of existing genetic data, i.e. from another testing company.
So I have a sort of motley crew of a dataset, but I’m optimistic it will make for an interesting and fruitful analytic substrate.
- Theme 2. Several developers built the tool they wanted for themselves.
This isn’t across the board, but I was struck at how many of the tools were born out of the developers’ own needs and desires. They had their own genetic data in hand and wanted to do something with it that existing tools couldn’t do, or couldn’t do in exactly the way they wanted. So with sufficient programming and bioinformatic skills, they build the tool they want and then expand it out for broader use. It reminds me of the phrase “If you build it, they will come.” Except here we have the developer saying something like: “If I build it, I will come…and then I’ll let everyone else who’s interested come, too.”
- Theme 3. The process of getting information is a source of information in and of itself.
I’ve been learning about tools by (1) studying the website and any associated papers or media coverage and (2) interviewing tool developers. The difficulty or ease of accessing these routes of information is sometimes telling about the tool itself. Some tools are run by companies where it’s very difficult to tease out any proprietary details of how they work (analytical tools and resources used, e.g.). Other tools are built on an idea of openness and are accordingly incredibly transparent about how they work, how many people are using it, etc. Reaching out to tool developers for interviews is similarly illuminating. But not always in the way I expected. Not surprisingly, several people are either ignoring or declining my inquiries, and it’s fair to say more companies fall into that category than not. But some companies are talking to me. And on the flip side, some academics are not. My goal is to get interviews with at least half of the tools I’m studying, which would be ~50% “response rate.” And the goal with qualitative interviews such as these is not to get to a statistically significant number, per se, but rather to gain some depth and texture to my understanding of who made these tools and why. Stay tuned for more!
In the opening scenes of Peter Pan, Peter has been separated from his shadow, and he breaks into the Darlings’ house in his efforts to get it back. Wendy is awoken by the sound of the chase and finds Peter, having gotten hold of his shadow, trying to reattach it with a bar of soap. Knowing this will not do, Wendy helps Peter to sew his shadow back on with needle and thread. The reunion of Peter with his shadow thrills and consoles him so much that he breaks into (at least in the musical version) the “I’ve Gotta Crow” number.
I was reminded of Peter Pan’s desperate search for his shadow a few weeks back after attending a lecture by Geoff Bowker, a prominent social scientist and scholar of Science and Technology Studies. During the talk, Bowker posited that we as individuals do not exist outside of our “data flows.” To unpack that a bit: our existence is bound up in our technology and our movements are incessantly tracked, analyzed, and commodified by the tech giants of the world: Apple, Google, Facebook, Amazon, etc. If you erase all the data points we leave scattered behind us like so many breadcrumbs as we move through our day, are we still there? Perhaps not, he ventured. Since the lecture, I have been thinking that because “big data” is such an integral part of us, such a constant feature of the traces we leave, that perhaps it has become like our shadow: our data shadow. Like Peter, maybe we are lost without it.
Existence outside of data
This was a rather disturbing idea to me — and based on audience questions, to some others in the lecture hall as well. My intuitive sense is that of course I exist outside of my data flows, detached from my data shadow. I’m signaling my age here, but I did not get an email address or start using the Internet until I was in high school. I didn’t join Facebook until about 10 years ago. Heck, I only joined Twitter in August. Did I not exist before I embedded myself into these webs of social media? Before I started trafficking in the Internet of Things? Not only do I intuitively feel like I exist to myself outside of data flows, I know I exist to my friends and family as well. So maybe it’s true that I don’t exist to Amazon outside of my online purchases or Prime Video streaming, or I don’t exist to Google when I’m not logged into Chrome – but they don’t define me to me or to those I interact with in the physical world.
Transparency without control
Back to the Bowker lecture…if we accept that we are at least in part defined by our data flows, what are we going to do about it? How do we fight back? (Supposing that, to my dismay, using Google “in cognito” windows falls short.) Part of the way to fight our way out of our data prisons, Bowker argued, was through transparency. If the algorithms companies use to crunch our data and mine our data flows for precious ore — if those can be made public, we will have made one step further towards liberation, towards political action, he argued.
Ok, I get that — maybe transparency is a necessary first step. But it is far from sufficient. Seeing what controls and potentially manipulates you may be a bit empowering, but that bigger agent still has the data, still builds and runs the algorithms. You need skills, training, and access to be able to do anything with or about the algorithm. It’s akin to when you’re on a web page and you know you can browse the source code with a right click and “view source.” But if you don’t know html code and CSS (cascading style sheets) and what not, you can’t really do much to change what you’re looking at or what it does. Right there, me not even being able to list all the things you’d need to know — that’s an example of my ignorance and inability to take control just because something is made visible to me.
Spoiler: here’s the point in my post where I draw in my dissertation research on consumer genomics, and in particular what people are doing with the “raw” or uninterpreted genetic data. I suspect that what motivates some people in this context is Bowker’s attractive idea that transparency alone is empowering. Just let me see my data – my sequence of A’s, C’s, G’s, T’s – and maybe that will bring me some insight and perhaps even some control over who I am, what diseases I may face later in life, etc.
But to the extent that might happen, I think it’s much more likely that rendering the raw data visible, downloadable, parse-able, may not do much at the individual level. Now granted I work with genetic data in a research context rather than in a clinical or medical context, so that skews my perspective. But the types of data people can currently get from consumer genomic testing, at least, is arguably much more valuable in the aggregate than to the individual who just has his or her data file. It’s analyses that are run on large scale genetic datasets that can teach us more about how human health and disease work, not so much about how DNA variants shape an individual’s trajectory.
I’m not saying transparency, of big data, algorithms, or genomes is bad — as Bowker indicated, it may indeed be a necessary first step. But let’s not get distracted by the allure of just seeing behind the curtain, of peering into Narcissus’ pool of data about ourselves.
There has been a lot of election “post-mortem” talk about living in bubbles. Urban bubbles, academic elite bubbles, blue state bubbles — all out of touch with and perhaps at times dismissive of rural America, no college degree America, red state America. [For a concise articulation of the problem, see November 8th New York Times editorial here.] I am likely guilty of all those bubble accusations: I live in Seattle so, blue state — check, urban center — check (though granted I grew up in Tennessee!). I have a Master’s degree and am pursuing a PhD. My graduate studies are in genomics, and in particular consumer genomic testing. Are my academic interests and pursuits elitist and out of touch with reality? Perhaps. But not entirely. Let me defend why.
Genomics in a bubble: the prosecution
First, why is genomics in general part of the bubble? An illustrative moment came during a talk I heard at the American Society of Human Genetics (ASHG) Annual Meeting back in October. I attend ASHG regularly and it is typically populated, as you would expect, with genomics cheerleaders (imagine the cognitive dissonance of devoting your career to something you don’t believe is important). But this particular presentation was about trying to implement genetic testing in low resource settings, specifically at Federally Qualified Health Centers. In the study being presented, providers at FQHCs described the challenges of using genetic information to guide care of patients with so many more pressing needs and problems: job, food, and housing insecurity among them. Genetics — in particular mildly predictive genetics that only influences complex disease risk by a smidgen up or down — does and should take a back seat when people are in need of food, shelter, and employment.
Genetics in the clinic aside, direct-to-consumer (DTC) genetic testing may be in even a more bubblier bubble. Indeed, empirical studies of DTC users have found them to be more white, more educated, and older than the average adult in the US. It is a luxury to have one or two hundred dollars kicking around that you can use to send off for your 23andMe or Ancestry DNA spit kit. It’s sometimes called “recreational” genomics for a good reason: people are often doing it for fun, or maybe even because someone gave it to them as a gift, and so why not. It’s also a luxury to have the free time to sit down at your computer and pour through your results on the company website and an even greater luxury to have the time to download your “raw” genotype data and poke around with it in third party interpretation websites (the specific area of my dissertation research). Yes, this is probably not your average person.
Genomics in a bubble: the defense
So we’ve established that it is probably not your average person who spends their time and money on DTC genetic testing and subsequent self-directed analysis in third party interpretation systems. But who is this person? What types of people are doing this and why? What are they doing with the information they get back and with what consequences? I’m trying to understand just that. To characterize this pursuit, however bubble-wrapped these people and these experiences may be.
My defense is that I do think this type of endeavor extends beyond just consumer genomics, and by understanding it better we can apply those insights to other types of pursuits. People, regardless of location, income, education, political leanings, etc., are typically engaged in some type of search: for personal and/or familial identity, for meaning, for connection with others, for health and wellness. Not everyone turns to their genetics for this, but people do turn to or seek out something. I think the dynamics of consumer genomics may well apply to these other areas. What motivates people, how do they satisfy those motivations, and what do they do with the result. That’s why I’ve chosen this area of research that I hope extends beyond the narrowly scoped instance of consumer genetics.
Genomics back in the bubble: precision medicine
Note I’ve been focusing on consumer genomics here, as that’s the subject of my dissertation research, but I do want to mention another hugely important, bubble-relevant application of genomics: genomics in clinical care, sometimes called “precision” or “personalized” medicine. As do many people, I fear that the integration of genomics in medicine may further widen existing disparities in access to prevention, care, and treatment in our health care system. That is, like many other medical technologies and knowledge, people with more education, higher SES, and better insurance are more likely to have access. Many genetic tests are not even currently covered by health insurance (it varies widely based on context, type of test, and insurer), meaning those who benefit are those who can pay out of pocket. Indeed, given limited resources and more pressing needs, genetics takes a deserved back seat, as we were reminded of by the ASHG presentation described above.
Carving out my corner of inquiry
So yes, I realize my doctoral work may be esoteric at times — narrow in focus at best and myopic at worst. But dissertation projects typically do (and must) carve out some narrow area of inquiry that may, on the face of it, be of concern to no one else but the student and their committee (and perhaps not even every member of the latter!). Part of that is just so we have a contained and sufficiently scoped project to complete in a few years. But we typically hope and strive to have broader relevance, to pierce the bubble and bring some goodness and understanding to others. And of course to take this time to listen and learn from those with different backgrounds, experiences, and interests. Including those who don’t buy that studying consumer genomics could ever teach us anything of real importance. Though I would hope to prove them wrong.
I recently spent 12 days vacationing in Italy with my mother and two older sisters. While my body is still processing large quantities of delicious cheeses, pasta, and gelato, my mind is digesting the experience of touring a foreign country with different norms, cultural nuances, and of course — a different language (though the diversity of head scratching bathroom set-ups also bears mentioning). On this trip, translation was always on the brain: translating my thoughts to others and in turn trying to understand the information presented to me, whether on signs, at train platforms, or spoken by short-tempered wait staff. Because despite my half-hearted attempts at learning Italian with the DuoLingo language app, or my high school courses in the nearby romance language French, I was nearly useless in speaking or understanding Italian.
The Need for Translation
My week plus of translation needs in Italy got me thinking about the role of translation in biology and, in particular, in genetics. In both contexts, translation carries at least two main functions: (1) operationalizing and (2) meaning-making. Operationalizing means to make functional, or to make a thing do. Translation is a key term in the Central Dogma of Biology. DNA isn’t terribly useful just sitting all spaghetti’d up in our cells. Rather, DNA carries instructions on how to make proteins that build us and do most all the work in our body (this is DNA as the “instruction book” or “blueprint”). The Dogma states that DNA gets transcribed into RNA, a molecule very similar to DNA but more easily accessed by other cellular machinery. Then RNA gets translated into protein, going from a nucleotide code (the A’s, C’s, G’s, and T’s – actually U’s, for RNA) to a chain of amino acids that gets all folded up into a beautifully complex protein. Translation is the operationalizing of DNA, the process that makes it do.
The Central Dogma is great and all but it’s a process scientists have understood for about half a century now, so not exactly breaking news. The challenge currently facing genetic researchers is truly understanding what different variations in genetic sequences actually mean for people’s health and well-being — and perhaps their identity. Here the challenge is translating knowledge of DNA sequence into actual meaning. Perhaps into meaning for an individual patient and their health care provider making a treatment decision. Or perhaps meaning for a large group of people by better understanding how a disease or other biological process works. The questions are more than just what changes in DNA do to proteins, which could take us back to that literal translation step of the Central Dogma. The questions spiral out: only ~3% of our DNA codes for proteins, but all that non-protein coding DNA could affect other things like regulation or as yet undiscovered cellular processes. Also, our genetics interact with other things in our bodies and in our lives, further complicating the meaning-making part of the translation puzzle.
The Tools of Translation
My needs for translation in Italy were pretty much the same: to be able to do things and to make meaning. I am not an expert traveler nor linguist, but I did have some amateur tools at my disposal. First and foremost: on my smartphone, Google Translate (with Italian downloaded for offline use) and an Italian phrasebook app. Off the screen, my sisters and mother who had also done some DuoLingo lessons, and my occasionally useful knowledge of French. Google Translate, which I used quite frequently, would often give me incomplete information — sometimes a word wouldn’t translate, or it would give me something I had no idea how to pronounce (and the audio pronunciation isn’t available offline). I knew some of the rules, for example: “ch” is a hard C, as in chianti, while “ci” is the “chuh” sound, as in ciabatta bread. But usually I was moving through the world with partial information, still enjoying myself and interacting meaningfully with my surroundings.
I bring up the amateur aspect of my translation experiences in Italy because I see parallels with the phenomenon of consumer genetic testing. While scientists are still wrestling to make meaning of human genetic variation, consumer companies have gone ahead (some would say prematurely) to make interpretations of personal genetic data available directly to consumers. The majority of these consumer genomics customers are, like I was with Italian, not specially trained to interpret or filter genetic information. Yet if given some tools and some rules, they can probably navigate the unfamiliar territory with some degree of enjoyment and success. Sure they might make a wrong turn or get caught in a tourist trap pizzeria (darn you Piazza della Signore in Florence!). But should they be denied access for their lack of expertise, or for being only armed with some amateur and partial tools of comprehension?
Of course in my Italy metaphor the answer is “No!”, but I recognize that consumer genomics is more complex — and newer, which makes it harder to identify and weigh potential risks and benefits. Should access to personal genetic data be limited to specialists? Should specialists make better tools to enable amateurs to pursue their own translational and meaning-making activities? Tourists have been bumbling around foreign countries since there was bumbling to be had: that’s just part of the human experience. Is bumbling around our own genomes also going to become part of the human experience?
Before I was geeky about science I was geeky about words. For my 16th birthday, my best friend gave me the “Encyclopedia of Etymology” — a giant tome about the origins of words (not bugs, people! That’s entomology). So of course I get excited when science and language interact, which happens a lot with metaphor. I even did my Master’s research thesis about metaphor (more on that later). One of the most surprising things I learned early on in that project was that most metaphor is actually lurking beneath the surface of how we talk and think on a daily basis, rather than being mostly confined to speeches and fancy poems (e.g., “Shall I compare thee to a summer’s day?”).
An example of a quite basic metaphor is that up is good and down is bad. Would you rather have things “looking up” or to be “feeling down”? Granted this metaphor may not hold across cultures, but in Western societies it is so ingrained as to almost be invisible. Note I did not discover all of this, but rather was introduced to these ideas in Lakoff and Johnson’s seminal 1980 book “Metaphors We Live By”. Think of Lakoff and Johnson like the Watson and Crick of modern metaphor studies. (If there is a Rosalind Franklin out there in this analogy, then my apologies in advanced for the omission!)
Metaphors for “big data” – h/t to Sara Watson
Metaphor is subtly sprinkled throughout our daily speech, and it can have powerful effects on how we think and act. Which is why it’s so important to identify metaphor and understand its sway on us. So I was pleased to recently come across self-proclaimed “technology critic” Sara Watson’s article on dominant metaphors for big data. She does a lovely job of breaking down dominant industrial metaphors for big data and suggests that replacing them with embodied metaphors, those more tied to our lived experience — our physical bodies — might help people exert more control over data and its downstream uses. Otherwise big data becomes this inevitable industrial, machine complex bearing down on us, so better hop on board or get out of the way.
Today’s society has a borderline morbid fascination with big data, which I’ve also written about previously in “Big Data, Big Deal?”, and you can see how the dominant metaphors perpetuate this fascination. A particularly problematic metaphor in my mind is that of data as a natural resource that should be mined, extracted, and purified. In this construct, data are commodified and spatialized. Just think of all the untapped reserves of “raw” data waiting for the boldest and most pioneering person to tap into: data logged daily by our smartphones, our Facebook profiles, and even our very bodies. In this metaphor, data become pre-factual and given, rather than contextual and imagined (whereas in actuality you have to conceive of something as a data point before you collect it — aha, even there, I did it: “collect data” as if I was picking wild huckleberries on a mountainside…which I recently did, incidentally). But full circle back to etymology: the very word “data” is from the Latin verb for “to give”….so it’s not totally our fault that it’s easy to take data as “a given.” (More on other cool things you can learn about the word “data” in my earlier post.)
The need to tease out metaphorical concepts
Sara Watson’s article articulates metaphors as “metaphorical concepts”, or “X is Y”: e.g., “Data is a natural resource.” Formulating metaphor this way is helpful in understanding the consequences or “entailments” of the metaphor and to raise further questions. If data is a natural resource, is it a renewable one or something finite (e.g., fossil fuel) that we may run out of? If data is a natural resource, who is “mining” it and who is using or buying it?
Metaphorical concepts are rarely stated outright, but identified through analyzing different expressions of the metaphor. You can see these expressions listed under the heading of the metaphorical concepts in Watson’s article: words like “raw,” or “trove”. Analysis of metaphor involves picking out those instances and then drawing out the underlying metaphorical concept.
Critique of a CRISPR metaphor analysis
Metaphor analysis that stops short of articulating metaphorical concepts is less useful. Last fall I wrote a piece along with two of my thesis committee members critiquing a metaphor analysis of the gene-editing system CRISPR that had this very problem. We argued that failing to articulate underlying metaphorical concepts resulted in a missed opportunity to understand who uses CRISPR to do what? Is CRISPR, as a technology, the subject of the metaphor or is the scientist using CRISPR the subject? It’s an important question of who or what has the agency to act and make decisions about gene editing.
Also, because the authors didn’t identify metaphorical concepts, most of the metaphors they report were about the genome itself rather than about CRISPR. It would have been easier for them to draw robust conclusions about CRISPR metaphors if they’d been able to separate out genome metaphors (to separate the “text” from its “editor,” as we allude to in the title of our critique).
Metaphors about genome sequencing: my MPH thesis
Oh – and did I hear someone ask about my Master’s thesis? I’m going to assume that’s a “yes.” For my Master’s in Public Health degree in Public Health Genetics, which I completed Spring 2014, my thesis project was a metaphor analysis of research participants discussing whole genome sequencing. I was fortunate enough to have access to several transcripts from previously conducted interviews and focus groups where people were asked to discuss genome sequencing in the context of research and medicine. No one was asked about metaphors specifically, but because of the frequency of underlying metaphors in spontaneous speech, instances of them popped up often in the participants’ discussions.
One of the most common metaphorical concepts I identified was “Genetic information is a weapon.” In some cases, getting personal genetic information was seen as a weapon in the hands of the individual, something empowering them to act, to defend themselves against disease or other potentially negative experiences. For other people, the weapon metaphor was one where genetic information was used as a weapon against them, to knock them over or leave them “shell shocked.” So even the same metaphorical concept can have different consequences, here depending on what or who is in control of the information.
Full disclosure was that initially I wasn’t forming my results as metaphorical concepts (“X is Y”) but more like keywords or domains (as we later critiqued in the CRISPR metaphor analysis). My committee member and resident metaphor expert, Leah Ceccarelli, strongly encouraged me to find the metaphorical concepts. My only real objection was “that sounds hard” (remember I’d never done formal metaphor analysis before), so once I realized that was lame I made myself do it – and ended up with a much stronger project for it.
You can read my whole thesis on ProQuest: search for title “Mapping Metaphor: A qualitative analysis of metaphorical language in discussions of receiving exome and whole genome sequencing results” (or, if you don’t have access to ProQuest, I’m happy to email it!). I also had peer-reviewed journal article published here. (Yes, it took an extra ~18 months to have that paper see the light of day – see my earlier discussion of the iterative and often trying nature of scientific publication here.) Meanwhile, here’s a table summarizing the main metaphorical concepts I identified.Table of metaphorical concepts from my thesis research project, with one or two example quotes from focus group and interview participants.
|Metaphorical concept||Example quote(s)|
|GENETIC INFORMATION IS A TOOL||[Getting genetic information] “might just be one additional piece of information to add to the toolbox”|
|GENETIC INFORMATION IS A WEAPON||[Receiving genetic results for a child] “could be a piece of information for them…to have in their arsenal for decisions that they’re going to make in their lives”
“So you don’t want too much information and, and with, I think with this, it’s so much. Genetic, there’s so much out there, you don’t want to be bombarded either.”
|GENETIC INFORMATION IS LIGHT||[Receiving positive results, e.g., about athletic ability] “would be like hey there's a light in the end of the tunnel”|
|GENETIC INFORMATION IS DARKNESS||“To know that I would develop early onset Alzheimer’s or, or something like that, I think it would be a consistent cloud over my life”|
|GENETIC INFORMATION AS GOODS INSIDE A BOX||“I’m going to want to [get] results on all of them. I’m curious like that. But I’m…not very confident. Kind of like opening Pandora’s box, do you want to know what’s inside?”
[On choosing when to receive results] “I want to open that box that’s, that’s mine.”
|GENETIC INFORMATION IS A PICTURE||“I don’t think I’m closed out to anything. I, I like the good and the bad because it all makes the whole picture.”|
|GENETIC INFORMATION IS A DOCUMENT||“If there was an architect going through the neighborhood and they were drawing plans, I want a copy of the plans of my house… I’m not going to build a house, I just want it.”
“…it would be nice to know, I guess I’m thinking of credit score like, here’s your credit score and here’s how you can improve it.”
Other recommended reading:
Ceccarelli, L. (2013). On the Frontier of Science: An American Rhetoric of Exploration and Exploitation. Michigan State University Press.
Condit, C. M. (1999). The meanings of the gene: public debates about human heredity. Madison: University of Wisconsin Press.
Lakoff, G., & Johnson, M. (1980). Metaphors We Live By. University of Chicago Press.
Through numerous conversations I’ve had with scientists, ethicists, and health care providers over the past few years, I’ve picked up on an odd and seemingly contradictory view of consumer genomic testing: it is both meaningless and dangerous. Not to paint a picture of complete professional consensus, as there is none, but from what I hear it’s these two threads that keep intersecting: danger…and irrelevance. To my mind, to be dangerous means to have power or at least be of some import, which implies having some meaning. This leaves me scratching my head and wondering: “Can consumer genetics really be both?” So I’ve been thinking of different scenarios that could explain these seemingly contradictory stances, which I explore below.
Recap of Consumer Genomics
First as a reminder, consumer genomics, or “over the counter genetics,” as I called it in a previous post, refers to companies offering genetic testing direct to the consumer (DTC), versus through a health care provider. These companies may return a range of reports, including on features such as genetic ancestry, who else in the company databases you might be related to, and risk for certain diseases. In addition, and of particular interest to my dissertation research, most DTC companies also offer to customers their “raw” (or un-interpreted) data for download.
You can see several ways both risk and unimportance could stem from these sorts of results. With genetic ancestry, you can learn roughly what proportion of your genome derives from which different geographic populations.
- Risk: the information is imperfect, because the reference populations are contemporary proxies, not the actual ancestral populations, and not all populations are represented.
- Irrelevance: people often already have a good sense of this (was I really surprised to learn my genome is 98.2% Northern European?), and even if they didn’t, is it really going to change most people’s conception of their racial, ethnic, and/or cultural identity?
With disease risk it’s hard to generalize given the range of diseases and the relative importance of inherited genetic variation to each one. But let’s focus on common, complex diseases such as type 2 diabetes or heart disease.
- Risk: people will not understand the limitations of the test results, which are about susceptibility and NOT diagnosis or deterministic prediction, and either over engage in or fail to engage in healthy preventive behaviors or screening tests. Or just generally freak out (“psychosocial distress”).
- Irrelevance: for many diseases, genetics plays such an infinitesimal role compared to factors in our environments that in the base case (i.e., barring some overwhelming family history) it’s usually pointless to even talk about inherited genetic factors.
Before I get into two scenarios that could explain the meaningless/dangerous tensions, for comedic relief here is one of my favorite XKCD comics on the topic. You can easily see the danger of misinformation and misinterpretation:
One should never avoid chocolate without solid evidence.
Scenario 1: Dangerous because it’s meaningless
The first scenario I want to explore is that consumer genomics is dangerous because it’s meaningless. This could be dangerous in two ways. First, it could be just a waste of time and distraction from more important things, especially when it comes to health. Indeed, in the early days of DTC genetics many experts worried that DTC customers would glut the health care system, making unnecessary appointments with their doctors to follow up on meaningless results, sucking up scarce time and resources. Customers (who are also potential patients, in this narrative) think a result is important, take it into their doctor, who in turn doesn’t think it’s important (and maybe rightly so).
And years later there appears to be some merit to that concern, as some studies found that DTC customers did increase the number of screening tests. (For a good review of empirical research on DTC customers and what they do and don’t do, see Roberts and Ostergren, 2013).
A second way meaningless could be dangerous is just the larger issue of people wasting their time and money. This isn’t a problem for the health care system, exactly, but a broader societal issue. Though to this I would argue that there are numerous other areas where we are not particularly protected against misdirected (or misspent) attentions. (I’m looking at you, Netflix!)
Scenario 2: Dangerous because of future meaningfulness
An alternate scenario is that consumer genomics is dangerous because it’s relatively meaningless now but hopefully won’t be in the future. That is, given our current knowledge of how genetic variation contributes to health and disease, there’s not much useful to be learned from the types of tests DTC companies offer. But, as research continues and we get better at integrating genomics with other types of information, the tests may improve. The whole idea of precision medicine is banking on this process getting better.
So there’s a bit of a reputation issue at stake. If people get exposed to genomics through DTC testing, where the results seem dubious and hand wave-y, they may not take it seriously 10 years down the road when their doctor wants to run a whole genome sequencing test on them. If genetics is currently recreational for most people, it may be difficult to recast it in a more serious light in the future. I’m reminded of a comment I heard at a genetics conference a few years ago, surveying the current and future state of the field of genetic research and medicine. The attendee made a comparison to a historical split between astronomy and astrology — at one point in time, everyone was just studying the stars, then the real science split off from the pseudoscience. He posed to the audience: do we (the professional genetics community) want genetics to go the way of astronomy (presumably what real researchers are doing) versus astrology (presumably the speculative activities of consumer genomics)?
It’s an interesting comparison, but also one that highlights the subjectivity of meaningfulness. Some people read their horoscope every day and find it quite meaningful; others would think that a superstitious waste of time. At they very least, I do think we need to keep an eye on the increasing commercialization of genetics and other types of health data, lest the potential gravity and utility of those data for research and medicine become obscured. And, if you do decide to do a DTC test, make sure it’s your spit and not your dog’s that gets sent in. 😉
I was recently sitting across from my financial adviser, at his desk on the floor of a busy bank in Seattle. I panicked as I realized that, through a slippery stream of acronyms and jargon, I had lost track of the conversation. ETFs, A-shares, C-shares, rights of accrual…I had even studied up on mutual fund terminology for this meeting, and yet I had still gotten lost. It was distressing, as I had always been a good student and was trying to be a good adult. Then I had two saving thoughts: (1) I am still a smart person, just not well-informed in this particular area and (2) I might not actually care.
This got me thinking about the role of acronyms, terminology, and general jargon in other areas. Including science. Genetics is a great one for jargon: DNA, WGS, SNP*. As genetic information becomes more widely available to non-genetics experts, the barriers to understanding put up by terminology become even more problematic. Or do they?
Disciplines and professions use specialized terminology (and yes, acronyms!) presumably for one of two reasons. First, sheer convenience. It’s too much work to spell everything out or give long winded definitions each time a concept is needed. Instead we use shortcuts. Within the given discipline or profession, this is generally unproblematic because the meaning is known and therefore the shortcut is sufficient to communicate the idea. (Side note that assuming shared meanings of terms and concepts presents a real challenge — and opportunity! — in interdisciplinary work.)
The second potential reason is for exclusivity. Terms and acronyms can be used to exclude non-experts from the conversation. The terminology allies someone with a discipline or profession, identifying them as a group member and as a practitioner/follower of a certain set of ideas, principles, and knowledge. It also carves out who doesn’t belong in that group. Just like when your older sibling used to speak in Pig Latin with her friends to keep you out of the conversation.
Need to know basis
My financial adviser is very kind, and I don’t believe he was intentionally excluding me from the conversation — but that was the ultimate effect. I could have stopped and asked him to break down and define each term, but that would have taken hours…and, remembering my earlier thought, I didn’t entirely care. I was content to be on a “need to know” basis about these transactions, and entrust him to make the best decisions on my behalf.
It was a wake-up call for me to realize that many people probably feel the same level of disinterest and “happy to defer” attitude about genetics as I do about my mutual fund investments. And probably this is a good thing, to have allocations of expertise so that we don’t all have to become an expert in everything. Those who would argue that genetic information should only be available through a physician, rather than direct-to-consumer, might subscribe to this idea. There’s no harm in trying to self-educate, but just because you have the internet at your disposal doesn’t mean you can study up enough to make fully-informed and autonomous decisions about every aspect of your life. Maybe getting genetic information about yourself should involve an expert fluent in all the jargon. In particular because you might not actually care enough, or have adequate time to study up.
I’m still not convinced, though, and think that personal genetic information needs to be made accessible to non-experts. Direct-to-consumer genetic testing companies do a lot of customer education, partly because no one is going to buy a product of which they have zero understanding. There is also a small but admirable set of genetic counselors out there who for decades have been working on educating patients and families to make well-informed decisions about getting and acting on personal genetic information.
For me, the breakdown with the mutual fund comparison is that my money, while personal, is far less personal than my actual body – my personhood. With intimacy comes the desire to stay informed, and to make autonomous (or at least partly autonomous) decisions. My health, well-being, and genetic information are much more intimate than financial investments, so I’m — well — much more invested in making those decisions.
*DNA=deoxyribonucleic acid, the molecule that carries genetic information in most organisms. WGS= whole genome sequencing, the process of determining the entire DNA sequence of an organism (e.g., a person!). SNP=single nucleotide polymorphism, a change in a single base pair of DNA. Pronounced “snip”
I was recently introduced to the term “seamful design” which, in contrast to “seamless design,” refers to a way of making things that doesn’t cover up all the messy inner workings — doesn’t remove all signs of the makers and their processes. A seamful design is one that may be more transparent, perhaps making the designers/creators more accountable to their users and audiences.
While my impression is the term has been used primarily in computer science, I got to thinking about seamful design in the process of scientific research. One of the most important products in science is the peer-reviewed journal article. These publications are often how a researcher’s scientific merit is judged and are a big part of hiring, promotions, and reputation-building in general. The genre is basically a write-up of the scientific procedure behind any study or experiment: you review what’s known, describe your questions, describe your methods, and then describe what you found.
But articles are atrociously seamless when it comes to accurately portraying how science is done. Articles are neat, linear, and nicely packaged. Research is messy, iterative, and linked to many other ongoing projects. And they take a friggin’ long time to get published. These realities are barely visible to the reader, and as a junior researcher I find this highly discouraging. After reading a paper I’m often left with the question: “But how did they really DO that?” Or, to paraphrase one of my professors: “What did they actually do on a Wednesday morning?”
The loooong road to a paper
In a nod to seamful design, and because it is somewhat cathartic for me, I will briefly walk you through how one paper I’m about to have published actually got done. Rewind to January 2015: I had just finished some analyses for my job and was presenting them to our group. Note the results were from analyses I’d started several months before, and we’d decided to tack on some extra checks to make sure everything was working well. One faculty member came up to me after the presentation and said, “Wow, those are really interesting findings — you should publish them!” (referring to the results of the extra checks).
I was flattered but didn’t pick back up on the idea until a few months later, as I had other more time-pressing work and school tasks to attend to. First we had to get our idea approved by a publications committee. That happened in March. In May I was finally able to sit down and start drafting the paper. By mid-summer I had a preliminary version to show my immediate supervisor. We went back and forth on it, and discussed who else should be a co-author. By autumn we had sent a draft around to the chosen co-authors, who had some minor suggestions and revisions. All along we’d been thinking about what journal to submit our paper to. There are considerations like prestige or “impact factor,” likelihood of acceptance, and turnaround time. We decided on a target journal and submitted in December 2015.
Two months later we got a set of comments from three anonymous reviewers of the article. To fully respond to their comments we had to run some additional analyses, some of which we didn’t really agree with but it’s the gesture that counts. The results didn’t tell us a whole lot, but since we’d already done the work and we thought it would appease the reviewers, we added the extra results in as supplementary material (like an appendix to the paper). The journal had originally asked for the revision back in 4 weeks, but we asked for an extension because I again was having to work on it amidst various other work and school tasks (note to self: next time avoid needing to ask for the extension! embarrassing…). We submitted our revision in April and then heard June 1 that our revision was accepted for publication. In a month or two the advance version of the paper will probably be available online through the journal website, but it will likely be several months more before it makes it into a print copy. So almost two years after I did the work, the paper will come out.
That account surely doesn’t show all the seams, and granted it was a much smoother process than it could have been. Often the first journal you submit to returns the paper without review. And other times your revisions get rejected so you’re back to the drawing board with finding another journal, and you’ve lost months trying to get into the first choice. I’ve also only been talking about the publication process, less about the stages of designing and conducting the research. But I just wanted to impress how little of the struggle and setbacks (and time!) the finished journal article can portray.
Showing the seams elsewhere in life
I’ve been talking about the genre of the scientific manuscript as an example of not very seamful design. But you can expand the argument out to how we interact with and perceive each other more broadly. For instance, on social media we see these selective, highly abridged versions of each other: a series of Facebook posts, Tweets, or Instagram pictures. It’s what we want to show and put out there, our best bits. Where are the seams? To be sure we shouldn’t advertise every misstep and pain, illuminate every wrinkle and pore. But we’re definitely not getting the whole story, just like a scientific paper doesn’t give you a very accurate picture of what lays underneath. We’re still left asking “But how did they DO that?”