About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Back in Court With Ariad and Lilly | Main | Airplanes and Chemicals »

August 9, 2006

Ray Kurzweil's Future

Email This Entry

Posted by Derek

Ray Kurzweil's people sent me a copy of his book The Singularity is Near quite a while ago when it first came out. I kept meaning to write about it, but several things kept interfering. One of the things was the book itself.

I'm of two minds about Kurzweil and the worldview he represents. As many will know, he's about as much of a technological optimist as it's possible to be, and I have a lot of that outlook myself. But I wonder - does it extend to my own field of research? More generally, and more disturbingly, am I only optimistic about the areas whose details I don't know very well?

These questions came up again when I read a recent op-ed by Kurzweil in the Philadelphia Inquirer. It's a good summary of his thinking, and it includes this paragraph:

"The new paradigm is to understand and reprogram our biology. The completion of the human genome (our genetic code) project three years ago is now allowing us to do that. This process is also exponential: The amount of genetic data we are able to sequence (decode) has doubled every 10 months, while the price for decoding each gene base pair drops by half in the same time frame (from $10 per base pair in 1990 to less than a penny today). For example, it took us 15 years to sequence HIV, yet we sequenced the SARS virus in only 31 days, and can now sequence a virus in just a few days."

That, to me, is a mixture of accurate information, reasonable optimism . . .and unreasonable assertions. Yes, we're sequencing things faster than ever before, and part of that increase comes through computational advances, which are a ferocious driver of everything they concern. But it's a very long leap from that to saying that such sequencing is allowing us to "reprogram our biology". Reading the DNA letters quickly does not, unfortunately, grant us an equally speedy understanding of what they mean. And we shouldn't forget that sequences are only a part of biological understanding, a realization that the genomics boom of the late 1990s drove home very forcefully and expensively.

Then we come to this:

"Being able to decode the human genome allows us to develop detailed models of how major diseases, such as heart disease and cancer, progress, and gives us the tools to reprogram those processes away from disease. For example, a technique called RNA interference allows us to turn unhealthy genes off. New forms of gene therapy are also allowing us to add healthy new genes. And we can turn on and off enzymes, the workhorses of biology. Pfizer Inc.'s cholesterol-lowering drug Torcetrapib, for example, turns off one specific enzyme that allows atherosclerosis, the cause of almost all heart attacks, to progress. Phase II FDA trials showed it was effective in preventing heart disease, so Pfizer is spending a record $1 billion on the phase III trials. And that's just one example of thousands of this "rational drug design" approach now under way."

Oh, dear. Let's take these in order. First, being able to decode the human genome does not allow us to develop detailed models of how major diseases progress. It allows us to begin to think about doing that, and to be, for the most part, mistaken again and again. Many diseases have a genetic component, or two, or a thousand, but we don't understand them yet, nor their incredibly tangled relationships with development and environment. You'd think we'd know the genetic components of diabetes or schizophrenia, but we don't, and it's not for lack of trying. And as for the diseases for which the genetic component is less important, the sequencing of the human genome has been a non-event.

And yes, there is a highly interesting technique called RNA interference which can turn "unhealthy genes" off. It works quite well (although not invariably) in a glass tube or a plastic dish. A plastic dish, that is, in which you have carefully cultered cells in which you have carefully determined the presence of the gene of interest. And for many interesting conditions, you first need to find your gene, for which see above. Moving out of the cell culture labs, it should be noted that RNAi has significant hurdles to overcome before it can do anything in human beings at all, and may (like its forerunner, antisense DNA) still be destroying venture capital twenty years from now. Readers of this site once voted it the currently hyped technology most likely to prove embarrassing.

As for new forms of gene therapy allowing us to add healthy new genes, well, that's another hope that I'd like to see fulfilled. But there have been a number of disturbing and fatal complications along the way, from which the whole gene therapy field is still trying to recover. For Kurzweil to leave that sentence in the present tense, in the sense of this-is-happening-right-now, is putting it rather hopefully.

And yes, we can turn off enzymes. Some of them. This has nothing to do with gene sequencing or RNA interference, though, or any other particularly new technologies - enzymes as drug targets go back decades, and enzyme inhibitors as drugs go back centuries. Of course, you need to find your enzyme and make sure that it's relevant to the disease, and find a compound that inhibits it without inhibiting fourteen dozen other things, but that's how I earn my living.

And yes, Pfizer hopes to make all kinds of money off torcetrapib, but I'm not aware that they used a "rational drug design" strategy. In the industry, we tend to use that term, when we can use it with straight faces, to mean drug design that's strongly influenced by X-ray crystal structures and computational modeling, but I don't think that this was the case for torcetrapib. Kurzweil seems to be using the phrase to mean "drugs targeted against a specific protein", but that's been the dominant industry mode since the days of bell-bottoms. And if there are thousands of programs comparable in size torcetrapib, they must be taking place on other planets, because there's not enough drug development money here on Earth for them.

Finally, the end of the paragraph. Where does all this lead? Later in Kurzweil's article, he says:

"So what does the future hold? By 2019, we will largely overcome the major diseases that kill 95 percent of us in the developed world, and we will be dramatically slowing and reversing the dozen or so processes that underlie aging."

And here, I think, is where I can clearly differentiate my thinking from his. As opposed to a pessimist's viewpoint, I agree that we can overcome the major diseases. I really do expect to put cancer, heart disease, the major infections, and the degenerative disorders in their place. But do I expect to do it by 20-flipping-19? No. I do not. I should not like to be forced to put a date on when I think we'll have taken care of the diseases that are responsible for 95% of the mortality in the industrialized world. But I am willing to bet against it happening by 2019, and I will seriously entertain offers from anyone willing to take the other side of that bet.

Why am I so gloomily confident? For us to have largely overcome those conditions by 2019, odds are excellent that these new therapies will have to have been discovered no later than 2014 or so, just to have a chance of being sure that they work. That gives us seven years. It isn't going to happen.

So I'm back to wondering: am I a technological optimist at all? I must be, because I still think that science is the way out of many of our problems. But am I only optimistic about things of which I'm ignorant? That's probably part of my problem, yes, painful though it is to admit. Am I willing to be as optimistic as Ray Kurzweil? Not at all. . .

Comments (69) + TrackBacks (0) | Category: Drug Development | General Scientific News


1. qetzal on August 9, 2006 8:59 PM writes...

One wonders if Kurzweil really believes that 2019 timeline, or merely judges (rightly) that it will bring him lots of attention.

Saying "we'll conquer them eventually, but it's gonna take a lot more time" just isn't very sexy, however accurate it is.

Personally, I doubt we'll overcome all those even by 2059. Some of them, probably, but not all.

Permalink to Comment

2. bonzo on August 10, 2006 12:27 AM writes...

Derek - Thank you! Every time I read or hear something by Kurzweil about biology I walk away convinced...that he's a very, very smart guy who doesn't know much about biology.

Permalink to Comment

3. Cherrie on August 10, 2006 4:08 AM writes...

Hey! I 'flipping' agree with you!!!

I sometimes think this sort of 'optimism' is bad because then the general public get frigtened about scientists 'playing god' and that sort of thing --> which never bodes well.

P.S. I love your blog - I only recently discovered it! =)

Permalink to Comment

4. lone electron on August 10, 2006 5:33 AM writes...

I can read the face of a playing card. It does not mean that I can win at Poker.

Permalink to Comment

5. daen on August 10, 2006 5:39 AM writes...

Kurzweil is, admittedly, the de facto leader of the more wildly optimistic tribe of transhumanists, and that optimism tends to translate to somewhat unrealistic timelines for certain technological developments. That said, all the right things are happening, not necessarily in the right order or at the right pace, to tackle diseases such as cancer and diabetes and schizophrenia. Consider that it's 53 years since the structure of DNA was determined, 31 years since Sanger developed the chain termination method of DNA sequencing, 21 years since Kary Mulis and colleagues developed PCR, six years since a draft of the human genome was published and three years since the human genome project has been declared as essentially complete. Companies such as Solexa and 454 Life Sciences now offer affordable, fast and accurate bulk DNA sequencing technologies, and existing players, such as ABI, are sure to respond soon. The current series of huge gaps in our knowledge between elucidating this wealth of DNA sequences and targetting disease pathologies will succumb, in time, to joint advances in technology, sufficient cash and the sort of dedicated cross-disciplinary enterprise that people like yourself bring to bear on these problems, Derek. In time.

Permalink to Comment

6. Industry Guy on August 10, 2006 6:11 AM writes...

2019.....hmmmm...I better get my retirement plans in order since we wil all be out of work by then....

Permalink to Comment

7. daen on August 10, 2006 7:07 AM writes...

On a not unrelated note, there's a story in C&EN (vol 84, number 31, p 17) about the clinical applications of systems biology in which Leroy Hood at the Institute for Systems Biology predicts that within 10 years DNA-sequencing technologies will be rapid and inexpensive enough that everybody's genome can be sequenced, with significant and positive implications for disease management. I was expecting the article to round off with a caution on timelines:

Hood is reluctant to put a firm timeline on his predictions about systems biology fulfilling its clinical promise, but he is sure that it will be sooner than anybody expects. "If you want to think about the future, you have to take into account the exponential change of technology," he says. "I've made a lot of predictions throughout my life. Inevitably, my predictions have been too conservative."

I guess he hasn't spent much time with Ray Kurzweil, then ...

Permalink to Comment

8. Anonymous on August 10, 2006 7:22 AM writes...

In some ways, Kurzweil is less optimistic than the National Cancer Institute - they are hoping to eliminate 'suffering and death from cancer by 2015'...

Permalink to Comment

9. RKN on August 10, 2006 7:48 AM writes...

I'm not familiar with Kurzweil's writing, but this flavor of unbridled optimism appears to be fed by the gene-centric view put forth by Dawkins and others that finally we've discovered the "source code." Accordingly, we can now "re-program" the machines as we see fit. It's a little difficult to know if people like this are merely sensationalizing the metaphor to sell books, or if they actually believe this stuff is practically true. A little time spent in the lab would go a long way toward sobering them up.

Giving me ten millions lines of source code to a complicated computer program, by itself, tells me next to nothing about the behavior of the system it encodes. I must first compile and link the code (transcribe & translate), and then run the system to have any hope of gaining a real understanding of how it works. Sometimes you can change the behavior of the system with one or two edits to the source code, but far more often it will eventually cause the system to crash, and usually unpredictably. Sound familiar? It is well known how drugs intended to produce a single change, for instance, those that target one or two receptors, very often produce a cascade of "phenotypical" surprises. This is why I think the focus in biology, and understanding disease in particular, is turning toward a systems biology approach. It pretty much has to.

Biologists are just now beginning to understand what software engineers have known for years.

You're wise to be skeptical of Kurzweil's claims.

Permalink to Comment

10. tom bartlett on August 10, 2006 7:52 AM writes...

I like Kurzweil's idea of uploading our brains into computers to cheat death. It seems more likely to me than solving all the multitudinous failings of our pathetic "intelligently designed" bodies. But I ain't holdin' my breath.....

Permalink to Comment

11. Opti Predictor on August 10, 2006 8:27 AM writes...

Here is a prediction that will make Kurzweil's circa 2019 prediction pretty accurate:

Once gene based therapies become proven in the lab - and some of them will be wildly successful at a very early stage - many "high net worth" individuals approaching old age will be willing to spend whatever it takes to get them. The result will be an FDA-avoidance scheme involving offshore clinics in soverign nations that are able to offer these services. Once the wealthy are cured at high cost, it is only a matter of a short period of time before the masses can afford it (true of every tech innovation - just look at plasma TVs).

Once this happens the FDA's analysis and approval process will be turned updide down and we'll see a revolution in health care.

I think 2019 is reasonable - within a four year window.

We tend to think linearly. We can't grasp the true power of the acceleration of knowledge, combined with AI.

Permalink to Comment

12. Justin B on August 10, 2006 9:27 AM writes...

The part that struck me the hardest is "by 2019, we will largely overcome the major diseases that kill 95 percent of us in the developed world." Not because of his blind optimism, but because of the word developed. What of the un- and under-developed world? Think they don't have problems too? HIV rates are still very high in these places, reaching 35+% in Botswana. It is sad to think that hundreds of millions will not receive treatment because they are poor. In Kurzweil's world, this new technology is only for the haves, not the have nots. When he writes a piece about how this technology has alleviated hunger, poverty, racial and religous intolerance, and violence, then I will be VERY impressed. But until then . . .

I'm thinking that if I develop a cure for HIV, but only the uber-rich can afford it, I'm not sleeping well at night.

And yes, the drug industry is a business, and making money is first on their list. Our work is not cheap, nor are the drugs we develop. But how long will it take for the "drugs of 2019" (forget the plasma TVs) to trickle down to the undeveloped countries? 5, 10, 20 years . . . never?

Permalink to Comment

13. daen on August 10, 2006 9:50 AM writes...

Justin, in all fairness to Kurzweil, as you yourself point out, this is hardly down to his technology bias. But, anyway, you're not quite correct to blame the capitalist model. The economic machinery which creates this uneven playing field has also enabled charitable trusts to be set up to tackle diseases dominant in less developed countries with access to funding equal to or in excess of what government sponsored research can offer, for example, in the form of the Bill and Melinda Gates Foundation. In many ways, this is already representative of the positive aspect of Kurzweil's future - a man who made a fortune of billions of dollars from selling computer software over the last 20 years is dedicating most of it towards addressing orphan diseases.

Permalink to Comment

14. Monte Davis on August 10, 2006 10:08 AM writes...

Kurzweil's at the intersection of two tendencies:

1) The millennial tendency -- whatever the area of interest, we are On The Cusp of Something That Will Change Everything. (Sometimes stated as This Can't Go On.) By suitable choice of axes and scales, you can always position Right Now at a point where an exponential curve is taking off.

2) The IT-PC-Moore's Law tendency -- to generalize from one very exceptional confluence of science, technology, and business to others. How many times have you heard "If [technology X] had progressed as fast as computers, we'd have..."

Trouble is, genetic and proteomic "code" isn't at all like the IT architectures we've created, from bit-flipping hardware through instruction sets, assembler, and higher-order languages. There are intriguing parallels, but... organisms are software and hardware mixed, horribly gnarly, almost infinitely re-entrant, "spaghetti" beyond a coder's worst nightmare. A given gene sequence or an alpha helix or a methyl group does many different things simultaneously in many different temporal, spatial, and biochemical contexts within one cell, let alone one tissue or organ.

So understanding much of it, let alone moving wholesale to the engineering (AKA medical) stage where you can make one change and have some confidence that it will have one clean result, is... uhh... going to take a while, Ray.

Permalink to Comment

15. Justin B on August 10, 2006 10:16 AM writes...


Our capitalist society is what it is, for good or bad (mostly good, in my opinion). In my view, Kurzweil should have omitted the word "developed" if he is indeed a humanitarian and serious about eliminating disease. I think he is an obsessive nut, but I hope he is successful.

Permalink to Comment

16. James Douma on August 10, 2006 10:25 AM writes...

This seems to be a pretty common response of experts to the idea of the technological singularity. The general feeling that technology overall is moving amazingly fast and getting faster combined with a resistance to the idea of rapid advances in their own field. Its hard to avoid being conservative about your own field when you see the problems so clearly and technology sometimes seems to creep along while you struggle with hard problems. Also, when it comes to your own field you expect to be able to imagine the revolution in detail whereas you can avoid that expectation in fields which you know relatively less about.

But the experts are often very wrong about how their own field is progressing. I think Kurzweil brings up that 'sequencing the genome' example a lot because it is easy to explain and, in part, because it shows how wrong the experts can be. A few years before the genome sequence was complete, when 90% of the work was done but only 10% of the genome was complete, the majority of experts were not expecting it to complete on time, and yet it completed ahead of time.

There are a lot of examples of this phenomenon. You wonder why smart, informed people are systematically wrong about the progress of their own field, and systematically underestimate what can be done in the long term.

The author of this blog isn't expecting any breakthroughs. He is expecting constant, steady progress in a manner that it comparable to his recent experience in the field. When breakthroughs happen they are surprising discontinuities in the progress of a field. Nobody plans on breakthroughs. Nobody really expects breakthroughs. But breakthroughs happen a lot and they happen faster when you are uncovering more new data and finding more new ways to look at it. RNA interference didn't exist a few years ago because it hadn't been discovered. Today it is an extremely promising new tool, and there are an amazing number of other promising new tools that have also just been discovered. This flood isn't going to stop. There will be more amazing tools coming faster and faster in the next two decades. Some of them will synergize. All of them will uncover a lot more data and give us new ways to look at it. And somewhere in that mix, there will be some breakthroughs.

Permalink to Comment

17. Canuck Chemist on August 10, 2006 2:35 PM writes...

From the comments of professional commentators like Mr. Kurzweil, it's obvious that most people have a pretty poor understanding of just where we stand in terms of our comprehension of biology. My analogy is that after the sequencing of the human genome, we have the complete works of Shakespeare in our hands, but it's all in Greek. It's a start, but it's a far, far cry from actually understanding (and effectively controlling) all of nature's beautiful subtleties expressed in our DNA. And that's not even opening the door on the higher-order complexities of epigenetics, post-translational modification, gene silencing, alternative gene splicing, etc. We've made tremendous progress, but in some ways we're still like cavemen peering under the hood of a car.

Permalink to Comment

18. TWAndrews on August 10, 2006 2:37 PM writes...

I'm not familiar with Kurzweil's writing, but this flavor of unbridled optimism appears to be fed by the gene-centric view put forth by Dawkins and others that finally we've discovered the "source code." Accordingly, we can now "re-program" the machines as we see fit.

What we've actually uncovered is not even the machine code! If we're trying to use computers as an analogy, what the sequenced genome corrosponds to is the order of ones and zeros on a hard drive. Without a *lot* more information, that doesn't tell you very much.

For instance, the very first step in getting any meaning out of the data, might be to figure out which bits are related to the operating system. Assuming that you can find sets of bits which are "differentially expressed" in response to turning your computer on, you'll probably be able to figure out some of the properties of your operating system eventually (but without some pretty strong assumptions about what an operating system looks like, this could be *really* hard).

Once that's determined, you can probably find, which some variable degree of accuracy, which bits the operating system groups together. Congrats, you've discovered that you've got files on your computer. Eventually, and again, with a not insignificant number of tests, you'll probably be able to figure out how the operating system groups files into sets. Nice job, now you've got an operating system, directories, and files within them. You don't know which files corrospond to programs, data or images, let alone being able to determine which files to modify to make a desired change, but still, it's something.

Just to get here would be an enormous amount of work, and probably thousands upon thousands of tests, wrong turns and mistakes. This is about where I see us as being in biology. We've got the genome, some information about how those genes are expressed, and ideas about some phenomena downstream from that.

Additionally, to carry the analogy a bit too far, we've done a good job of finding some files which we know corrospond to various types of bugs, glitches and machine crashes, and found ways to modify these (mostly) without wreaking the system. But even so, we're along way from being able to really understand our biology, let alone reprogram it.

Basically Kurzweil's on something if he thinks that we're going to have everything figured out in the next 15 years. It's going to be this century's project to understand our biology and achieve the level of mastery over it that we developed over the pysical world in the 20th century through physics and engineering. Which is to say, profound, though hardly complete.

Permalink to Comment

19. Still Scared of Dinosaurs on August 10, 2006 6:26 PM writes...

"2019.....hmmmm...I better get my retirement plans in order since we wil all be out of work by then...."

Speak for yourself. They'll keep running trials for at least a few years after the last research lab shuts down.

;) I think...

Permalink to Comment

20. secret milkshake on August 10, 2006 7:08 PM writes...

In late 1940s, the pop-magasine prediction was that every family suburban house would soon have a tidy nice "atomic reactor" in the garage, about the size of a water boiler and the uranium fuel sticks would be obtained from the drugstore.
(I bet it was not a nuclear physicist who wrote this charming piece of prognostication.)

Permalink to Comment

21. mungojelly on August 11, 2006 7:49 AM writes...

Artificial intelligence is the crucial factor, the tremendous transformation. It's perfectly reasonable to think that we wouldn't be able to transform the world as quickly as Kurzweil & others are saying that it's going to-- we couldn't. We're just human beings. We are fantastically slow, ponderous & conservative compared to the cyborgs who are going to burn this place up.

We always want to think that it's going to stop here. There's been a bunch of transformations, but they're not THAT big a deal, really, are they? Now we can read genomes, so what. So what, indeed-- if that were the end. The point isn't that these transformations are going to change everything so completely; the point is that these transformations are only the mast poking above the horizon. Things will keep changing.

At the moment we have tools to look at genes, but we still mostly use human brains to develop hypotheses about that data. It's far far too much data for the tiny number of bipedal primates who do that sort of thing to fully comprehend. Agreed. What will change soon to make it comprehensible is the nature & practice of science itself. New kinds of understanding, nonhuman understanding, are emerging around us like flowers in spring.

Permalink to Comment

22. Derek Lowe on August 11, 2006 8:58 AM writes...

MJ, I'm not seeing it over here. There are billions of dollars waiting for someone who can apply some of that AI magic to predicting human blood levels and toxicities of drugs, to pick just one notable example, and no one's stepping up to claim 'em. Even a ten or 20 per cent improvement would turn someone into the Bill Gates of the drug industry. But it isn't happening, or at least not yet.

Permalink to Comment

23. markm on August 11, 2006 11:01 AM writes...

TWAndrews, great analogy. However, if I understand the current state of DNA research correctly, they have a basic understanding of the "file system": transcription start and stop markers have been identified, and mechanisms to patch a gene (a "file") together from several separate segments are partly understood, but there's a long way to go to a complete understanding of gene regulation (what controls which files get read).

But that also reveals more complexity - things like cross-linked files, that are errors in a human-designed OS requiring replacing the file from a backup, are apparently routine parts of the design in DNA. The same chunk of DNA can be read backwards, forwards, and from different starting points to produce different proteins. It's not to conserve DNA, because something like 90% of the DNA seems to be junk, not part of a working gene. It's like your hard drive was choked with cruft, including unrecovered parts of erased files, unused files, and pieces of a million outdated OS's, while you were using programming tricks with the last 10% of your space to use one bit for multiple purposes.

Or maybe all the DNA is there for a reason, and we only understand 10% of it.

Finally, the DNA isn't the program that runs the machine. It's the program that writes the programs (proteins) that build and run the machine.

Permalink to Comment

24. RKN on August 11, 2006 7:13 PM writes...


For the record, transcription always occurs in the 5' -> 3' direction, never "backwards," at least not to encode RNA from DNA. The enzyme reverse transcriptase does "read" RNA from 3' -> 5', but this is to create DNA from RNA.

Permalink to Comment

25. Erik Nilsson on August 11, 2006 10:54 PM writes...

In #23, markm defends a computer analogy for DNA.

Analogies for things people have made (such as a computer) for something in nature (such as DNA) are popular, and inherently misleading. Designed things like a computer are orderly and comparatively simple. Biology isn't like that. We have the same problem when we pretend that the wiring diagrams we make for metabolic and signalling pathways are like those for electronic circuits. But biological wiring diagrams are not design plans. They only map our incomplete and imperfect discoveries about the system.

There are other problems with comparing DNA to a computer program. One big one is that programs have a purpose, but biology does what it does, whether we can ascribe a purpose or not.

Perhaps a less inappropriate analogy would be, given the bits corresponding to a computerized accounting system, determine the structure and capabilities of the system, in total ignorance of GAAP and tax law.

I'm an optimist about the promise of biomedicine, but it seems like every time nature has had an opportunity to be elegantly complicated, nature has exploited that opportunity. Biotech isn't easy, and it won't get easy. Various cancers may be routinely curable in 10 years, but it will be a lot longer than that, before we can "reprogram" our bodies away from disease.

Permalink to Comment

26. TWAndrews on August 12, 2006 2:02 PM writes...

Analogies for things people have made (such as a computer) for something in nature (such as DNA) are popular, and inherently misleading.

Absolutely true. But given that people are going to make them anyway, I was just trying to point out that even if the Biology : Computers analogy was apt, we *still* aren't anywhere close to having the "source code" and being able to "reprogram" anything.

Permalink to Comment

27. daen on August 12, 2006 6:45 PM writes...

Analogies for things people have made (such as a computer) for something in nature (such as DNA) are popular, and inherently misleading.

Yes and no. Talking about "files" and "operating systems" and "source code" is simply too high a level of structure. DNA and its transcription machinery can, however, be considered as a real world example of a Turing machine, and can be analysed as such.

As a sidenote, I find it curious that Roger Penrose, author of the anti-strong AI book "The Emperor's New Mind" railed so strongly against the mind being algorithmically based (and hence, he argued, being capable of encompassing the analysis of algorithmic impossibilities such as the nature of halting states, which cannot be done with a Turing machine - and deriving elegant mathematical proofs), and yet failed to point out that the very machinery that builds the brain that houses the mind is dependent on essentially algorithmic processes ...

Permalink to Comment

28. Erik Nilsson on August 13, 2006 12:02 PM writes...

TWAndrews wrote: I was just trying to point out that even if the Biology : Computers analogy was apt, we *still* aren't anywhere close to having the "source code" and being able to "reprogram" anything.
Fair enough.

daen wrote: DNA and its transcription machinery can, however, be considered as a real world example of a Turing machine, and can be analysed as such.
This is literally false, and it's not a useful analogy. Neither DNA transcription nor RNA transcription is Turing complete. You can't compute with a strand of RNA and a ribosome. You just can't. DNA and RNA polymerase are likewise lacking computational properties. The two systems together aren't Turing complete either. Furthermore, transcription from DNA is not a general computational process.

DNA looks superficially like the cartoon version of a Turing machine. This cartoon is just a handy way to visualize the mathematics, so the superficial resemblance doesn't prove anything, but DNA transcription doesn't even look much like the cartoon. For example, ribosomes can't write values onto the RNA strand. It makes as much sense to say that a tape recorder is Turing complete, because there is a tape and a head. But a tape recorder isn't a computer, either.

Clearly, biology is able to peform computation-like operations. At the extreme case, since humans can simulate Turing machines in our heads (and since a human invented them in the first place), humans as a whole have Turing-like properties. But to consider "computation" on the molecular level, you have to drag in a lot more biology, and consider transcription regulation mechanisms. Plus, transcription regulation is often fairly important to "computation" (in the sense of a predictable response to sensory inputs) in prokaryotes, but what I've seen suggests it's a lot less important in eukaryotes like us.

This is one of the big problems I have with Kurzweil: he uses inappropriate analogies with computation to come up with laughable estimates for the computational power needed to accurately simulate cognition. Kurzweil should talk to more neurologists.

Permalink to Comment

29. daen on August 13, 2006 1:59 PM writes...

Erik, I must disagree with you, as you seem to be a bit confused here. What has Turing completeness got to do with specific Turing machines? I never claimed any attributes of general computational completeness for DNA and its transcription mechanisms, merely that they may be analysed as a type of Turing machine. Not all Turing machines write to the tape (although they may do) and not all of them have valid halting states (although they may do). In essence, the 64 codon "states" of groups of three nucleotides all simply cause the "head" to advance the "tape" and read the next "code", until it (hopefully) hits one of three stop states. Fairly dull, but still, I argue, a Turing machine, not just a simple tape recorder.

Permalink to Comment

30. The Chieftain of Seir on August 14, 2006 6:37 PM writes...

I am an ignoramus. But with that disclaimer out of the way...

From my reading of history(and I know, past result say nothing about the future) progress has happened far faster than people ever predicted. But all throughout that time, the Ray Kurzwil types have always been wildly wrong.

They were wrong because along with unanticipated progress comes unanticipated problems. The diseases that use to kill people off in the 1800's were largely cured by the 1900's. Only for people to start dying of things that never use to be much of problem in the 1800's. The cure for cancer was promised to us back in the 1950's. We are all still waiting. We know a lot more about cancer than we ever did and we can fight it far more effectively. But cancer has proven to be far more complex than researchers originally thought back in the 50's. And I am sure we can all think of more examples along that line.

In order for Kurzwil and his ilk to be proved right, we must have unexpected solutions to the problems that we now face and no unlooked for problems. But that is not how the human search for knowledge has worked up to this point. We always seem to be discovering problems that we never dreamed existed along with our solutions to old problems.

I expect to see unexpected solutions, but I also expect unexpected problems to come along.

Permalink to Comment

31. Erik Nilsson on August 15, 2006 4:54 PM writes...

daen wrote I never claimed any attributes of general computational completeness for DNA and its transcription mechanisms, merely that they may be analysed as a type of Turing machine.

I'm glad we agree that DNA transcription is not a universal Turing machine. But in that case, all you are cliaming is that a Turing machine can be described that simulates DNA transcription. That's a very weak claim, since a Turing machine can be described that simulates any deterministic mechanism. For example, there exists a Turing machine that simulates a simple padlock, but that doesn't mean a padlock has interesting computational properties.

You made your claim with regard to Turing machines in an attemt to rebut or qualify my argument that analogies between DNA and computers are misleading. From your clarification, it appears we both agree that such analogies are misleading.

Transcription isn't nearly as simple as the cartoon version, of course. Incidentally, the Turing machine you outline in your response appears intended to simulate transcription from mRNA into proteins, not any process directly involving DNA, but the machine fails to simulate essential characteristics of this mechanism. Most obviously, the machine does not model protein production, which is a central aspect of transcription. You also are restricted to a single ribosome per mRNA strand, which is not the usual situation.

What you've shown is that the popular, incorrect conception of the Central Dogma resembles the popular, incorrect conception of computer science. That probably says something interesting about how our brains work, which probably explains why a smart guy like Kurzweil can say such silly things about computers and biology. That was my point. You say we disagree. Where?

Permalink to Comment

32. Brooks Moses on August 15, 2006 4:56 PM writes...

Heh. This is an interesting perspective for me, given that most of what I know about pharma I learned here, but I have had a small bit of experience with artificial intelligence.

If anything, Kurzweil's preditions about medicine and biology seem (from your descriptions) to be less off the mark than what I've seen of his predictions and claims about artificial intelligence. I distinctly remember some years ago sitting around the lab with a couple of AI researchers, looking at some new "AI" interface that Kurzweil had put up on his website, and roundly mocking it.

The thing about MJ's "artificial intelligence" predictions, Derek, is that the necessary artificial intelligence to do this kind of thing is something that exists "right around the corner"; nobody is claiming that much at all of it exists now. The whole core of Kurzweil's predictions is that this corner is really close, and that as soon as we get around it, things will change very dramatically because the AI will be in a position to bootstrap itself without requiring laborious decades of human hard work. The expectation is that it will be a discontinuity, such that current experience with the capabilities of AI for things like drug design will not be valid predictors of what's possible after the discontinuity.

Frankly, from what I've seen, the idea of a discontinuity like that is ... well, on historical timescales a thousand years from now, it might look like a discontinuity, but I expect it will happen on a scale more like a century or so, and getting to that corner is going to take at least that long. If it even happens that way at all, which I am also dubious about; AI research has created a substantial quantity of things that turn out to be useful but not anything like what we expect strong AI to be.

Permalink to Comment

33. daen on August 15, 2006 7:28 PM writes...

Erik, you seem to be getting rather bogged down with cartoons and padlocks and details, and are still, it seems, deeply confused about my intent and, it seems, some of the actual molecular biology. I did not mention simulation, nor proteins, nor ribosomes, nor the Central Dogma, nor computation, nor translation. If you read it again, my last comment tries to draw an analogy between the action of transcription (ie RNA polymerase acting on a single DNA strand) and the actions of Turing machine.

Permalink to Comment

34. Erik Nilsson on August 16, 2006 11:20 AM writes...

OK, so you like an analogy between Turing machines and DNA. That analogy proves nothing and illuminates nothing, mostly because it's a bad analogy.

I am not confused, but I'm bored. Feel free to have the last word if you like.

Permalink to Comment

35. celebrim on August 16, 2006 11:53 AM writes...

"Derek - Thank you! Every time I read or hear something by Kurzweil about biology I walk away convinced...that he's a very, very smart guy who doesn't know much about biology."

Funny, every time I read or hear something by Kurzweil about artificial intelligence I walk away convinced... that's a very, very smart guy who doesn't know much about computers.

Given his demonstratable successes in that area, that may seem a counter-factual claim, but I'm not so certain that it is. I think that his successes rather proof that he's made achievements in certain extremely narrow areas. I'm increasingly convinced that this does not translate to any broad understanding of anything else. Kurzweil is the living proof that idiots can be brilliant, and geniuses can be idiots.

Either that, or he's just selling something to suckers.

Permalink to Comment

36. TallDave on August 16, 2006 11:59 AM writes...