I got a lot of interesting responses to my post about DIYbio and how modeling innovation in biotech on computer hacker culture may lead to a science that is less "democratized" than what is being proposed. My friend Adam pointed me to Jaron Lanier's work criticizing the "open" and "free" culture movements online as both unfair and leading to cultural stagnation. While I don't agree with all of Lanier's arguments about the prospects of an open digital culture, he makes a lot of really important points that resonate with my feelings about the future of science based on the open online model, in particular for synthetic biology. He addresses synthetic biology specifically in his book You Are Not a Gadget, but it is his discussions of the difficulty in building large software even as hardware improves exponentially according to Moore's Law, the lock-in of stultifying software standards, and the economics of the cloud that are particularly interesting and valuable to a discussion about the future of synthetic biology.
On synthetic biology he writes that "wikified" biology, science that breaks down the already loose institutional barriers between individual scientists and between individual species--as genetic sequences are passed between people who put them into different organisms--will limit both biological evolution and technological innovation. He argues that cellular boundaries around early genetic sequences are what drove evolution out of the primordial ooze, and by breaking down the importance of those cellular boundaries, that encapsulation, we lose the locality that drives evolutionary novelty. He says the same of academia:
Academic efforts are usually well encapsulated, for instance. Scientists don't publish until they are ready, but publish they must. So science as it is already practiced is open, but in a punctuated, not continuous, way. The interval of nonopenness--the time before publication--functions like the walls of a cell. It allows a complicated stream of elements to be defined well enough to be explored, tested, and then improved.
Here I agree that academic science is already quite open, although perhaps not open enough in some cases. If you publish a paper that includes genetic constructs that you built, you are required to send that genetic material to anyone from an academic lab that asks to use it for research purposes. I have never had a problem getting something I needed from labs in the synthetic biology community or from anywhere else. The ultra-secretive nature of some labs before publication, however, can certainly be detrimental. I have friends who aren't allowed to present their work at department lab meetings, for fear of being scooped by colleagues down the hall. This overprotective, fearful environment holds back the students who can't get any outside feedback on their work, and can hold back genuinely collaborative scientific progress.
At the same time I don't want to build off of work that hasn't been vetted or proven in some way (which doesn't necessarily have to be publication). A totally open repository of genetic parts, as it basically exists now in the form of the synthetic biology Parts Registry, therefore can have a lot of problems. The registry has countless parts that are essentially nonfunctional, but you can only tell after considerable time wasted searching for a part, finding and getting it, and then sequencing, testing, and verifying part functions. This work does improve the quality of the registry, "wikifying" biological data collection by outsourcing quality control to unpaid users, but we still won't necessarily approach the quality of parts made and maintained by individual scientists. By divorcing the part from its original context, the lab or scientist who built and tested it, we end up losing some of the value of the work that went into producing it, and we lose some of the ability for genuine collaboration. DNA (and even DNA plus detailed data characterizing its function in a specific lab) alone isn't necessarily enough for a creative and innovative project in synthetic biology.
Completely abstracting the functions of genetic material, environmental contexts, and species boundaries is dangerous as well, and I don't agree with Lanier that this is the future we're headed towards, if only because this is just not how biology works. Biology is powerful exactly because it doesn't work like computers. There is no standard way of doing anything. Genetic pieces are passed between bacterial cells (or even between food and the bacteria in our guts) but context and evolution within individual cells and populations matter for life. Interactions between genetic material and the cellular environment, and each cell with its ecological and "social" context creates the adaptable, evolvable, beautiful diversity we seen in the natural world.
By defining arbitrary standards early on in our understanding of how these genetic elements work in their rich biological contexts, and early in our ability to engineer novel functions, we lose sight of much of the complexity of biology and get stuck in what could become difficult and useless technological cycles. Indeed, the BioBrick standard as it was first defined a decade ago does not allow for proteins to be fused to one another in-frame, and uses restriction enzymes that are rare and expensive. BioBrick cloning is wonderfully convenient for certain applications, but enforcing a standard that doesn't take into account the realities of how biological parts are made and used in different contexts and for different projects is inefficient, which is why there are almost as many "standards" for pseudo-BioBrick cloning are there are labs in synthetic biology today.
Lanier warns against any movement to enforce a specific way of doing things, any locking in of standardized forms in technology development. By defining one industry standard, we may open up an easier way to make things at industrial scale, but we also lose diversity in how we think about and use whatever we're standardizing, making true innovation more difficult. He argues that this is particularly true in software development, where standards locked in during the early days of computing are difficult to throw off, particularly when designing and maintaining large-scale software packages and operating systems. Small programs are easy to make in new paradigms, but large programs are slow to change and extremely costly to manage and innovate, despite leaps and bounds in the speed of computer hardware.
The same can be said for synthetic biology, where small genetic networks with innovative but limited novel behaviors can be routinely and relatively easily made, but large-scale combinations of smaller networks or synthetic pathways with more than a dozen genetic components remain elusive. Many commentators on biological technologies claim that they are progressing even faster than Moore's Law, with the prices of gene sequencing and synthesis dropping precipitously every year. But this price drop does not necessarily equate to a similarly exponential ability to understand gene sequences or create complex new biological behaviors. Synthetic biology will not necessarily follow Moore's Law because human scientific creativity and evolutionary change are fundamentally different from how transistors work.
Creativity in synthetic biology designs, in novel syntheses of biological knowledge, bio-technical expertise, and engineering concepts are done by groups of individual hard-working people. These people and their work should be valued as something special, something that can't be done by simply increasing the number of base pairs of DNA being synthesized. So too should we value the power of evolved biological systems as something different from the designed electronic systems that inhabit our world today. What do we gain by trying to fit biology into the structures that have become locked into computer engineering? What would we gain if instead we created a new kind of engineering, one that centered on learning more from and about the biological world?
- Log in to post comments
I agree with you... I don't know much about biology but it's beginning to feel like brute force application of computer engineering paradigm won't quite work with living systems.
An entirely different engineering paradigm specifically for living systems is an attractive prospect, but where to start... Maybe protocell/mincell projects can come up with something useful. Do you have any ideas?
I think protocells are really interesting, and you could certainly create a new engineering paradigm from there, but as a biologist I'm more interested in starting from current living biological diversity. I'm interested in what cells can do now, after 4 billion years of evolution. How can we manipulate what exists in nature? What can we learn from this kind of engineering? Can an understanding of living cells allow us to better engineer non-living things--for more sustainability, efficiency, robustness?
Dear Ms Agapakis,
Your thoughts are enlightening, your style of writing super duper.
Check this out - ATP Synthase to Kanye West
http://www.youtube.com/watch?v=XI4oniz-h_Y
I made it, figured your the type that digs it too.
Cheers
-Andrew
There are infinite ways of mapping a biological system into an hierarchical network of interacting components, integrating atoms to populations of organisms with everything in-between, and in each case there will be *cycles* in the complete network.
My intuition tells me that these cycles would represent the physical and logical compartments that separate components that could react with one another. The membranes delineating these physical and logical compartments could be something obvious, such as a bona-fide phospholipid membrane, or something more abstract, such as a metabolic pathway like the carbon cycle, or even something as cryptic as a spatio-temporal logic circuit encoded in the cis-acting promoters of a concert of genes.
Each compartment has a defined interface: for the membrane, it would be all the particles that could cross it or send signals across it; for the metabolic network, it would be the upstream and downstream substrates and enzymes input and output from the reaction pathway; for the spatiotemporal logic circuit, it would be the spatiotemporal molecular signals that initiate it and that it outputs.
In each case, an interface boundary can be identified. It could even be redesigned. And in many cases, the functional cycles delineated by the interface ARE standard across many different organisms. So I disagree with your assertion that "there is no standard" way of doing anything in biological systems.
I think the standards are there. We just don't have the tools to characterize or engineer with them (where engineering means constructing a design without needing to experiment). Yet.
Protocells and synthetic organelles seem like an interesting place to start.
Of course there are modules shared across different species, modules and processes that can be understood, removed from their original context, and redesigned to work in a different cell type. This is how I make my living after all. Just because there are commonalities between cells across evolution doesn't mean there are "standards." Standards imply intentionality and stasis, the opposite of evolution, and in the study of biology they imply a stable definition of life, a decision made by some computer scientists about how biology should be done, and a calcification of our understanding of what biology is and can do. I'm not arguing against doing research in synthetic biology and trying to understand the modules and systems shared in biology through redesign, I'm arguing for grounding synthetic biology in a curiosity about biological processes, not in saying "this is how it has to be forever."
Every generation makes analogies between what they make and understand (technology) and what they don't understand (biology). People used to be treated like steam engines, and now like supercomputers. Maybe the analogy is more apt these days, but it doesn't mean that it's the only way to go. You talk about biology in your comment in terms of computation: cycles, interface, standards. Why can't biology be it's own (messy, confusing, wonderful) thing?
Is the notion of "stable evolution" or "intentional evolution" or even "engineered evolution" non-sensical, or is it consistent with the new paradigm of engineering tools synthetic biologists are necessarily going to develop to do biological engineering?
I certainly agree with you that new paradigms are necessary. I find the flow of society's favorite analogy over the last 150 years to be very interesting, from pressures and mechanical systems to electromagnetics, to computation, and most recently, to networks of all kind (esp. social). So I'll leave you with a quote you may already know, by Carl Woese, from A New Biology for a New Century:"
Aha, "so there you have it!" We need things that exhibit "(stable) complex, dynamic organization" (preferably on the macro scale) so we can make analogies to them.
Thanks for writing such a thought-provoking piece Christina!
I deeply agree with you that biology is its own wonderful, messy thing. But I don't think that establishing some standards is inherently impossible.
I think a field like synthetic biology has much to gain by finding consensus on what are good practices and sound design principles. If we consider synthetic life, I could easily imagine a kind of 'industry standard' in which a certain mode of converting information to molecules (like ribosomes) becomes the standard.
This doesn't have to stifle innovation! When such a standard becomes widely adopted and ingrained in certain designs, rogue and radically new strategies might be more rewarding. Consider Google, which introduced pagelinks as the standard form of currency for finding relevant pages. Google might have set a standard of how we find stuff, but since it has turned into a massive and slow company now, the door is opened for new players that are not hindered by bureaucracy, that can introduce other (more social) forms of finding search results, using Google as a 'template' upon which to build.
In a way, this is similar to evolution. Life builds (and becomes dependent on) previous innovations. Sometimes it innovates in a unexpected and wonderful fashion. Sometimes it keeps the robust and thrustworthy, because it works.
I think that standards will absolutely be useful for industry, but I think you're again relying on analogies to computers and computer companies that may just not be what biology will be like and should be like. There are standards in how people do experiments and how people think about science and those things gradually change as new innovations pop up, but when we talk about actually standardazing LIFE, that's where I think the problem lies. This is definitely more of a philosophical argument I guess than a business one, in terms of what life means and most importantly how we understand what life is and what it can do. By focusing so much on the development of standards, locking in ways of how we see and engineer life now with an incomplete understanding of what life can do, I think we miss out on a lot of possibilities and we miss the point of why engineering living systems is interesting in the first place.