Is Molecular Nanotechnology “Scientific”?

Pr­ereq­ui­site /​ Read this first: Scien­tific Ev­i­dence, Le­gal Ev­i­dence, Ra­tional Ev­i­dence

Con­sider the state­ment “It is phys­i­cally pos­si­ble to con­struct di­a­mon­doid nanoma­chines which re­pair biolog­i­cal cells.” Some peo­ple will tell you that molec­u­lar nan­otech­nol­ogy is “pseu­do­science” be­cause it has not been ver­ified by ex­per­i­ment—no one has ever seen a nanofac­tory, so how can be­liev­ing in their pos­si­bil­ity be sci­en­tific?

Drexler, I think, would re­ply that his ex­trap­o­la­tions of di­a­mon­doid nanoma­chines are based on stan­dard physics, which is to say, sci­en­tific gen­er­al­iza­tions. There­fore, if you say that nanoma­chines can­not work, you must be in­vent­ing new physics. Or to put it more sharply: If you say that a simu­la­tion of a molec­u­lar gear is in­ac­cu­rate, if you claim that atoms thus con­figured would be­have differ­ently from de­picted, then ei­ther you know a flaw in the simu­la­tion al­gorithm or you’re in­vent­ing your own laws of physics.

My own sym­pa­thies, I con­fess, are with Drexler. And not just be­cause you could ap­ply the same ar­gu­ment of “I’ve never seen it, there­fore it can’t hap­pen” to my own field of Ar­tifi­cial In­tel­li­gence.

What about the Wright Brothers’ at­tempt to build a non-biolog­i­cal heav­ier-than-air pow­ered fly­ing ma­chine? Was that “pseu­do­science”? No one had ever seen one be­fore. Wasn’t “all fly­ing ma­chines crash” a gen­er­al­iza­tion true over all pre­vi­ous ob­ser­va­tions? Wouldn’t it be sci­en­tific to ex­tend this gen­er­al­iza­tion to pre­dict fu­ture ex­per­i­ments?

“Fly­ing ma­chines crash” is a qual­i­ta­tive, im­pre­cise, ver­bal, sur­face-level gen­er­al­iza­tion. If you have a quan­ti­ta­tive the­ory of aero­dy­nam­ics which can calcu­late pre­cisely how pre­vi­ous fly­ing ma­chines crashed, that same the­ory of aero­dy­nam­ics would pre­dict the Wright Flyer will fly (and how high, at what speed). Deep quan­ti­ta­tive gen­er­al­iza­tions take strict prece­dence over ver­bal sur­face gen­er­al­iza­tions. Only deep laws pos­sess the ab­solute uni­ver­sal­ity and sta­bil­ity of physics. Per­haps there are no new quarks un­der the Sun, but on higher lev­els of or­ga­ni­za­tion, new things hap­pen all the time.

“No one has ever seen a non-biolog­i­cal nanoma­chine” is a ver­bal­ish sur­face-level gen­er­al­iza­tion, which can hardly over­rule the pre­cise phys­i­cal mod­els used to simu­late a molec­u­lar gear.

And yet… I still would not say that “It’s pos­si­ble to con­struct a nanofac­tory” is a sci­en­tific be­lief. This be­lief will not be­come sci­en­tific un­til some­one ac­tu­ally con­structs a nanofac­tory. Just be­cause some­thing is the best ex­trap­o­la­tion from pre­sent gen­er­al­iza­tions, doesn’t make it true. We have not done an atom-by-atom calcu­la­tion for the syn­the­sis and be­hav­ior of an en­tire nanofac­tory; the ar­gu­ment for nanofac­to­ries is based on qual­i­ta­tive, ab­stract rea­son­ing. Such rea­son­ing, even from the best available cur­rent the­o­ries, some­times goes wrong. Not always, but some­times.

The ar­gu­ment for “it’s pos­si­ble to con­struct a nanofac­tory” is based on the pro­tected be­lief pool of sci­ence. But it does not, it­self, meet the spe­cial strong stan­dards re­quired to cer­e­mo­ni­ally add a be­lief to the pro­tected be­lief pool.

Yet if, on a whim, you de­cide to make a strong pos­i­tive as­ser­tion that nanoma­chines are im­pos­si­ble, you are be­ing ir­ra­tional. You are even be­ing “un­scien­tific”. An un­grounded whim­si­cal as­ser­tion that to­mor­row the Sun will not rise is “un­scien­tific”, be­cause you have need­lessly con­tra­dicted the best ex­trap­o­la­tion from cur­rent sci­en­tific knowl­edge.

In the nan­otech­nol­ogy de­bate, we see once again the se­vere folly of think­ing that ev­ery­thing which is not sci­ence is pseu­do­science—as if Na­ture is pro­hibited from con­tain­ing any truths ex­cept those already ver­ified by sur­face ob­ser­va­tions of sci­en­tific ex­per­i­ments. It is a fal­lacy of the ex­cluded mid­dle.

Of course you could try to crit­i­cize the fea­si­bil­ity of di­a­mon­doid nan­otech­nol­ogy from within the known laws of physics. That could be ar­gued. It wouldn’t have the just plain silly qual­ity of “Nan­otech is pseu­do­science be­cause no one’s ever seen a nan­otech.” Drexler used qual­i­ta­tive, ab­stract rea­son­ing from known sci­ence; per­haps his ar­gu­ment has a hid­den flaw ac­cord­ing to known sci­ence.

For now, “di­a­mon­doid nanosys­tems are pos­si­ble” is merely a best guess. It is merely based on qual­i­ta­tive, ab­stract, ap­prox­i­mate, po­ten­tially fal­lible rea­son­ing from be­liefs already in the pro­tected be­lief pool of sci­ence. Such a guess is not re­li­able enough it­self to be added to the pro­tected be­lief pool. It is merely ra­tio­nal.