Functional composibility (i.e. can modularized “parts” be assembled” is a central challenge in synthetic biology, on which much of the overall enterprise depends. It is a challenge that bears on questions of design and composibility, e.g. to what extent are living systems suceptible to the classic goals of engineering such as standardization and modularization. It is also a challenge that bears on questions of ontology: what is being made in synthetic biology.
A question we have is: how does functional composibility play out in nanotechnology?
Tags: 12 Comments
12 responses so far ↓
just to provoke: is there such a thing as “the classical goals of engineering”?
Modularity is one among many problems in engineering, related to standardization and scalability… but is it the case that there is a consensus? I think synthetic biology benefits from being far away from actual engineering practice in that it can project this claim onto engineering. In nanotechnology, especially within schools of engineering, i don’t think this is the case.
Point taken regarding the generalization.
To put the question with specificity, then. In synthetic biology a principal question is: to what extent are living systems susceptible to modularization, abstraction, and standardization. The problem is functional composibility: assembling systems and getting the elements of the systems to function in specified and regular fashion.
The question we have is whether or not a parallel problem exists in nano. To what extent to challenges associated with assembling biological systems take form in the design and construction of non-biological systems?
Hi Chris, hi Gaymon. Chris, what do you mean by “synthetic biology benefits from being far away from actual engineering practices”?
So I think there are a couple of ways to approach this:
1) the problem of “functional composibility” or modular design is a hard problem in all realms of engineering, and the question you are actually asking is “what difference does the nature of the object make to the ability to standardize and stabilize function?” In this case the comparison to computer engineering is apposite. The discussion around the “software crisis” (ongoing since 1968) has always included questions about modularity (Michael Mahoney at Princeton and Nathan Ensmenger at U Penn have both written about this history). In Computer engineering the problem is that it’s so easy to “scale up” (produce many identical copies) that standardization becomes a hard problem. It’s so easy to change the function of something without cost that people do it all the time, and use it as a strategic advantage over competitors (microsoft word format, for instance, forces people to use microsoft word). Standardizing software therefore is a hugely difficult task. Chapter 5 of my book is about this, and focuses on the standardization of UNIX. Chapter 7 focuses on modularity in Linux and Apache.
2) I would try drawing a distinction between stablization and standardization. The former concerns the practical ability to make something stable, repeatable, and functionally identical. The latter concerns the Political (and economic) process whereby one attempts to force one kind of stabilized object to be used by everyone for a particular purpose. They aren’t temporally ordered. With sufficient legitimacy, the standard can come first (a so-called de jure standard) which sets the parameters for the stabilization of an object. But in synbio’s case, I suspect stabilization will come first, because there are probably some kinds of biological objects more easily defined and replicated vis a vis function than others, and this will drive what one can do with these parts. Also, stabilizing a function is different from stabilizing an object. In Nano, for instance, lots of people are working on the mechanical control of ATP as a motor. There are dozens of ways to work with it, and lots of demonstrations of how to control it, but not of these efforts are defined in terms of a specific function, such as it’s ability to lift, actuate, spin or something in a specific way.
3) Drexler/Endy vs. everyone else. In some ways, the comparison is again between the imagineers and the real work. Drexler’s big books of “theoretical engineering” which were the first nanotech books are filled with engineering principles and the ideals of creating carefully standardized molecular parts for the construction of new molecular manufactories. However, most, if not all, working nanotechnologists think he was full of it, and never went near a working a lab to really test the feasibility of these ideas. Some of them may turn out to be useful, but it won’t be a fully worked out engineering science of the sort he envisioned (and managed to convince lots of other people to envision). I suspect something similar is going on with Endy, but not quite as extreme… a kind of cart and horse problem, where he can clearly see what will have been needed to create synthetic biology as an engineering discipline, but he doesn’t want to wait for all the work of stabilization and experimentation necessary to get to that point. Both of them are crucial for the conceptual innovation… but it doesn’t map onto the practical problems.
Talia, I mean that the people involved in synbio are far away from engineering in a pedagogical sense. While there are people trained in engineering disciplines, like Tom White, it is not situated within a milieu where the problems of engineering are immediately present (such as in a mechanical, chemical or electrical engineering department or amongst engineering corporations). This has a couple of implications, I think: one is that there is no clear commercial outside to the endeavor which sets constraints on the practice, or to which synbio engineers would naturally look to in order to learn from. The other is that engineers in other fields do not necessarily feel any obligation to critique synbio as to whether it is getting things right or wrong as far as engineering principles are concerned, largely because it is its own cutting-edge thing, and not (yet) relevant to any other engineering practice.
Tom White is from Celera. Tom Knight is from MIT. He is definitely an engineer and has been engineers for a long time. Randy Rettberg spent his career at Intel. Randy runs the registry. Most of the MIT group are engineers.
They think Roe Smith provides the right analogies.
Some of the people out here are engineers as well but more in the chemical engineering side of things.
I think the question is on the “what is the object being worked on?” and how manipulable is it?
We had an excellent interview yesterday with the guy doing the registry down at JBEI and tomorrow another interview with one of the test bed leaders.
oops s/White/Knight/g
I stand by my assertions though– it doesn’t matter if engineers are involved… of course there are engineers involved. the point I’m making is that they are not doing this in an engineering milieu– or maybe the question I’m asking is “what milieu are they working within?” I don’t think it is engineering, which is all the more obvious if Roe Smith (the Civil War) is their point of reference.
Chris can you clarify further the point about engineers vs. engineering milieu. Many the principle players in synthetic biology are engineers, strongly identify as engineers, and work in engineering departments (Jay Keasling, Tom Knight, Drew Endy, Ron Weiss, Adam Arkin, Chris Anderson).
In response to the three points Chris made on functional composibility:
1. We wonder if the analogy to software development is the right one. It seems to us that the problem in that case concerns political economy. Political blockages are certainly in play in synthetic biology. However there is the additional problem of whether or not biological systems can be modularized and re-composed, even if all the political stars were aligned.
2. The question of the relations among stabilization, standardization, and function in synthetic biology hangs in part on the persistent question of what is a module. The engineers in SynBERC refer to modules as objects, functions, and processes.
3. It is not clear that Endy does know what it takes to make synthetic biology work as an engineering discipline. First there is the question of his analogies. Second, where he uses analogies, there is a question of whether or not he uses it right.
I think at an academic research level, people can call themselves whatever they want, especially if what they are doing is restricted to a single laboratory, or a series of labs. People end up in departments with wacky names for all kinds of reasons. Some departments of chemical engineering are very old school… some have transformed themselves into chemistry and molecular bio-engineering hybrids (like Harvard). In short, the names and affiliations don’t indicate the stability of a field of research.
There is, however, an aspect of engineering that should be highlighted (and the desire present in the synbio folks is to participate in this aspect): scalability and industrial production. What makes standardization appealing, it seems to me, is that it leads to larger and larger scale implementation of the kinds of things people can build with a lot of trouble and expense, by hand, in their labs. In this sense, synbio is not yet at this stage of engineering, nor is it emerging out of an existing industrial/large scale practice. It doesn’t emerge out of electrical engineering of semi-conductors or big pharmaceutical engineering. It may get swallowed up by those domains, but it isn’t currently operating within that milieu.
As a result, the attempt to standardize is largely a cerebral pursuit–and not, as it is in the case of standardization in say, telecommunications or automotive manufacturing, a political process. That milieu of standardization is not yet engaged. This is why I suggested that what they are trying to do are find ways to stabilize what they are working on.
Now, having said that, it is also clear that low-level standards regarding the kinds of things that the synbio people are making might also be of interest to folks at the ASTM, or ANSI or ISO… but it isn’t yet clear why– precisely because it is yet clear what these things will do.
In nanotechnology, there are extensive debates about materials, because they concern standardization of the nomenclature, standardization of the definition of the material in nano-scale as opposed to bulk, and definition of the process of producing them. These debates are occurring in the context of materials engineering, however, so all the questions people ask are of the form “is a carbon nanotube a different kind of material than a sheet of graphite?” because graphite is well defined already in materials engineering. So at least some realms (but not all) of nano are operating within well-defined engineering milieus that really set the terms of what constitutes a new object.
Another aspect of this milieu is the professionalization. Engineers, until the advent of information technology, have been a highly professionalized cadre, replete with licenses and exams and societies. An interesting question might be: what are the relevant professional engineering societies for this group of people?
I love it when you use the Royal We, gaymon
1. I think focusing on the *failures* of standardization in computer engineering is the right place to compare. It’s not clear to me that
information technology objects can really be effectively “modularized” in the way that the synbio people seem to take as uncontroversially easy to do in that domain. Software engineering has remained “pre-industrial” for 40 years now, never making the leap beyond the need for art and hand-craft in the creation of software. I think synbio would do well to reflect on the never-ending “software crisis” if it truly wants to understand how stabilization and standardization work… it’s a problem no one has been able to solve, and the result is that we live in a world of buggy, constantly malfunctioning, obscure and unmaintainable software. god forbid we should replicate that with biological engineering.
2. Whatever a module is, it is currently be defined in the absense of goals, it seems. A datasheet for a resistor or a nut and bolt may look just like a datasheet for BBa_F2620, but I don’t have a clue what the latter is for. I understand what it does, but I don’t know what system it fits into. When do I use it? Is this a cart and horse problem maybe? A Resistor reduces the amount of current flowing on a line, which I might need to do in a circuit system. A nut and bolt combines two parts together. What does BBa_F2620 do?