So You Want to Colonize the Universe Part 2: Deep Time Engineering

Part 2: Deep Time Engineering

(1, 3, 4, 5)


So, with “Gotta go Fast” as the high­est goal, and aware of the fact that the amount of com­pu­ta­tional re­sources and think­ing time de­voted to build­ing fast star­ships will ex­ceed by many or­ders of mag­ni­tude all hu­man thought con­ducted so far, due to the im­por­tance of it...

I set my­self to de­sign­ing a star­ship to get to the Virgo su­per­cluster (about 200 mil­lion light years away) in min­i­mum time, as a lower-bound on how much of the uni­verse could be colonized. I ex­pect the fu­ture to beat what­ever bar I set, whether hu­man­ity sur­vives or not (it turned out to be about 0.9 c)

Now, most peo­ple fo­cus on in­ter­stel­lar travel, but the in­ter­galac­tic travel part is com­par­a­tively un­der­ex­plored (see com­ments). We have one big ad­van­tage here, which is that we don’t need to keep mam­mals around, and this lets us have a much smaller pay­load. In­stead of de­liv­er­ing a ves­sel that can sup­port earth-based life for hun­dreds of mil­lions of years, we just have to de­liver about 100 kg of Von Neu­mann probes and stored peo­ple, which build more of them­selves. (The true num­ber is prob­a­bly a lot less than this, but as it turns out, it isn’t harder to de­sign for the 100 kg case than the 1 mg case be­cause there’s a min­i­mum vi­able mass for dust shield­ing, and we’ll be cheat­ing the rocket equa­tion.)

Be­fore we get into in­ter­galac­tic star­ship de­sign (part 5), I want to take a minute to point out the field of Deep Time en­g­ineer­ing, which is some­thing that I just crys­tal­lized as a con­cept while work­ing on this.

Note that what­ever star­ship de­sign you’re build­ing, it has to last for 200 mil­lion years, get­ting bom­barded by rel­a­tivis­tic pro­tons and dust the whole way, and even with rel­a­tivity speed­ing things up, you’re still talk­ing about build­ing ma­chin­ery that last for tens of mil­lions of years and works with ex­tremely high re­li­a­bil­ity the whole way. This is in­cred­ibly far be­yond what en­g­ineer­ing nor­mally does, it takes god-like lev­els of re­dun­dancy and re­li­a­bil­ity, and if you’ve got some­thing with mov­ing parts, there’s ero­sion by fric­tion to con­sider, and also 200 mil­lion years worth of cos­mic rays… I didn’t fo­cus on ac­tual solu­tions that much, but just the aware­ness of the ex­is­tence of tasks which re­quire build­ing ma­chin­ery that works for hun­dreds of mil­lions of years sparked some­thing.

Eng­ineers have shorter time hori­zons than you might ex­pect. In en­vi­ron­men­tal en­g­ineer­ing (my ma­jor), we typ­i­cally fo­cused on a 20-50 year de­sign life for build­ing wastew­a­ter treat­ment sys­tems. They’re also de­pen­dent on the elec­tri­cal grid for func­tion­ing. I think that I could de­sign a 500-year treat­ment plant that also wasn’t de­pen­dent on the elec­tri­cal grid. It would take a while, bring in quite a few non­stan­dard con­sid­er­a­tions, and be far out­side of the scope of nor­mal de­sign, and a bunch of stan­dard ap­proaches (like us­ing en­ergy-hun­gry air pumps to aer­ate the wa­ter) wouldn’t work. A plant that does this would also have an enor­mously larger foot­print than stan­dard wastew­a­ter treat­ment plants.

Sev­eral-hun­dred or sev­eral-thou­sand year solu­tions are in a very differ­ent de­sign space than stan­dard solu­tions.

I should also make the note that we’ve figured out how Ro­man Con­crete works, which is far more ero­sion-re­sis­tant than stan­dard con­crete (it lasts for sev­eral thou­sands of years, and is far more re­sis­tant to salt­wa­ter than stan­dard ce­ment), and this is why the Colos­seum is still stand­ing. Ba­si­cally, you just use sea­wa­ter in­stead of reg­u­lar wa­ter when mak­ing it. Also the steel beams in reg­u­lar con­crete which give it ten­sile strength in­stead of mere com­pres­sive strength ac­cel­er­ate cor­ro­sion sig­nifi­cantly. How­ever, reg­u­lar con­crete takes a few hours to cure enough to ap­ply weight, and cures fully in about a month. Ro­man Con­crete takes two years to fully cure. And this is why very few places use Ro­man Con­crete, even though it lasts over an or­der of mag­ni­tude longer. (I did find an ar­ti­cle about a Hindu tem­ple un­der con­struc­tion that was us­ing Ro­man Con­crete, and was de­signed for a thou­sand years, though).

Even in civil en­g­ineer­ing, the land of roads and bridges and build­ings, you tend to see 100-year de­sign lives at most, as well. I should note that there are ta­bles that tell you the av­er­age mag­ni­tude of a 100-year flood (largest flood ex­pected in 100 years), and these are used in de­sign. And also the teach­ers men­tioned that due to cli­mate change, ex­treme weather events are more likely to oc­cur than the ta­bles in­di­cate. But they didn’t ex­plic­itly con­nect these two things, it was left un­stated for the stu­dents to click to­gether, and there was also an un­stated im­pli­ca­tion that go­ing to the higher-re­dun­dancy sys­tems that’d han­dle 100 years+cli­mate change would lead to peo­ple ask­ing you why you’re us­ing 1,000-year flood num­bers in­stead of 100-year flood num­bers and the de­sign wouldn’t pass.

There are ex­cep­tions. The sea walls in the Nether­lands are sized for 10,000-year flood num­bers, and I got a pleas­ant chill up my back when I read that, be­cause there’s some­thing re­ally nice about see­ing a civ­i­liza­tion build for thou­sands of years in the fu­ture.

There’s also the at­tempt to de­sign nu­clear waste stor­age that warns peo­ple away for tens of thou­sands of years, even if civ­i­liza­tion falls in the mean­time. This pop­u­lar ac­count is worth read­ing, as a glimpse into long-timescale en­g­ineer­ing.

But in gen­eral, Deep Time Eng­ineer­ing is pretty un­der­ex­plored, be­cause it re­quires much higher costs, much higher re­li­a­bil­ity, a larger foot­print, and about all of the ma­chin­ery that you’d buy isn’t rated for hun­dreds or thou­sands of years, there’s no sup­port­ing in­fras­truc­ture for en­gag­ing in con­struc­tion pro­jects of that de­sign life.

The spe­cific man­i­fes­ta­tions of it would vary widely by field and the speci­fics of what you’re build­ing, but in gen­eral it seems to be a dis­crete Thing that hasn’t pre­vi­ously been named, and that our civ­i­liza­tion ne­glects.

Build­ing a 100-mil­lion year (or even billion-year) star­ship is an es­pe­cially ex­treme ex­am­ple of this. For my spe­cific star­ship de­sign, the only thing that ac­tu­ally re­quires con­tin­u­ously run­ning the whole time is the an­ti­mat­ter chilling sys­tem to get it to 0.1 K when the cos­mic microwave back­ground is 2.73 K (oth­er­wise it heats up enough that you lose all your an­ti­mat­ter to evap­o­ra­tion against star­craft walls by the time you get there). This takes less than a watt of power to do, but keep­ing an an­ti­mat­ter cool­ing sys­tem (and stor­age sys­tem, al­though su­per­con­duct­ing coils help im­mensely) con­tin­u­ously run­ning for ge­ologic timescales is a very im­pres­sive feat. Also, all the ma­chin­ery for de­cel­er­a­tion has to still work at the end of 100 mil­lion years of cos­mic ray dam­age and such, and there’s a part in there where end up firing a multi-gi­gawatt nu­clear en­g­ine for a few mil­le­nia to tar­get a spe­cific star, which is also go­ing to be ex­tremely hard to de­sign for that level of re­li­a­bil­ity. (imag­ine the ra­di­a­tion dam­age to the en­g­ine from that level of power, it won’t be pretty).

Re­pairing nanobots help, but it’s still go­ing to be an im­pres­sive feat.