• Google
    This Blog Web

October 2011

Sun Mon Tue Wed Thu Fri Sat
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          

RSS Feed

Bookmark and Share

Email Feed



  • Powered by FeedBlitz

« Accelerating Future | Main | The Discovery Machine »

December 20, 2005

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451db8a69e200d8342ea59f53ef

Listed below are links to weblogs that reference Watching the Meme Spread:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

michael vassar

I sincerely doubt this has most of the implications of MNT. It sounds like, at the very most, a substantial improvement in existing manufacture capabilities, but probably not even that. The fraction of current consumer spending that goes into the sorts of goods that such a device could produce is negligible, and the fundamental technology doesn't sound likely to be cheaper than existing factories.
It's a great toy for nerds, and might lead to the proliferation of new consumer products, or new variations on old products.
In a best case scenario, this plus the web will stimulate a burst of surprising inventions, and allow their rapid dissemination, but even that will need good software. My guess is that this will do for a variety of crafts and for simple inventing what computer graphics programs did for graphic art.
Here's a proposed crude scale of importance
fire, agriculture, and MNT are 10s, electricity, printing, and steam engines 9s, steel, radio, and flight are 8s,
automobiles, antibiotics, and cannon 7s,
fermentation, plumbing, and refrigeration 6s,
cinema, washing machines, and anesthesia are 5s,
the biggest blockbuster drugs, VHS, and nylon are 4s,
microwave ovens, typewriters,and home manufacturing of this sort are 3s
2s, like drain cleaner, teflon, or a new music format come out every year but impact our lives noticibly and often produce billions of dollars of wealth.
1s include most major successful inventions. The post-it note, velcro, the coat hanger, reduced fat ice cream that actually tastes decent, we're surrounded with them

Mike Treder, CRN

I sincerely doubt this has most of the implications of MNT.

I never meant to suggest that it did. Far from it. But getting people familiar with the concept, in outline form, is important.

michael vassar

Hmm. It looks to me like my '1s' produce on the order of $one billion of value, and each step above that involves 4-5 times as much value creation, though of course, some of that value comes in the form of GWP, some in the form of consumer surplus, and in a few cases, such as cars that number is lower than the market size, but reflects externalities. Cannon, being zero sum, don't really fit in this framework.

That gives MNT a value creation estimated at between $250 trillion and $20 quadrillion.
Probably about right for relatively early versions. Raise standards of living globally to somewhere between that of average Americans and that of the richest 5% of Americans, better insome ways, worse in others, less status, land, and ability to purchase services, better toys and physical comfort. That estimate, combined with the amount of money being devoted to MNT gives us some estimate of the efficiency of markets.

Another point, a fleshed out table of this sort, possibly with values produced by poll or vote and technologies more clearly defined, would be a useful tool for calibrating change. I'm of the impression that it would show results compatible to the Huebner results of declining per-capita innovation from WWI to the late 1980s (pretty much exactly the cold war) with a recent increase in the last 15 years, back to 1930s or 1940s levels.

michael vassar

Another insight that this categorization exercize inspired from me is this. MNT is revolutionary, but in it's early implementation, assuming we survive it's really not the biggest thing ever, just the fastest. It won't produce abundance, because we already have abundance, we just have a distribution problem. It might solve the distribution problem, if it's allowed to, by enabling communism without totalitarianism by making people no longer one of the means of production that must be controlled. However, crude MNT, prior to advanced nanomedicine, is much less transformative than the whole industrial revolution, taken as a single thing. I don't even think that it's as transformative as the computer revolution taken from Univac to pre MNT nanocomputing. Nanomedicine is, however. Nanomedicine is much much bigger than MNT. Bigger than anything to come before. Than everything before combined. On the scale I described, it might be a 12 or a 13. In so far as it enables AI and singularity, it's easily greater than a 50. This is a reminder to keep our eye on the ball. If we can get the benefits of nanomedicine without MNT, that's what we should aim for. By contrast to nanomedicine (life extension and mind/body modification), MNT is a trivial poison bauble, a risk to the species and to human freedom for a prize that is, in the scheme of things, trivial. I'm not calling for relinquishment. That's impossible. I'm not calling for an end to planning. We should plan, and MNT still looks like the best path to nanomedicine to me. But we should remember, nanomedicine is the goal, MNT a terribly risky means to an end.
Oh yeah, and that 50+ value, for AI? If you are really able to think about it clearly, to take that seriously, and to do something useful about it, keep that in mind. Nothing, absolutely nothing on earth or in the heavens above is worth increasing the risk of that going wrong by the slightest imperceptible jot. Not saving the rain forests... not your personal integrity... not avoiding a billion year Reich.
Nothing At All.
Of course, the hard part is to take things that seriously and not go crazy. I don't even recommend trying to take it seriously until you know enough history and evolutionary psychology to see where traditional morality comes from. Enough to notice on your own that not a single assasination in the history of the world seems to have ever made the world one jot better and understand what that shows about the mental states within which assasination is possible, (Kazinsky f*cked that one up but good). Getting AI right, by which I mean doing it safely, is not just the most important thing ever, it's close to being the most difficult thing ever, and dabblers make success less likely not more. Before you think about trying to act directly, please read this (forgiving the writing style)
http://www.singinst.org/action/seed-ai-programmer.html
Or, if you prefer, ignore what I just said and focus on nanomedicine and on surviving MNT. Excluding the singularity this is still far and away the most important thing in the world.

Mike Deering

If you think we already have abundance, you clearly don't understand abundance, at least the kind of abundance that is possible with MM. While nanomedicine isn't the only benefit of MM, I'll admit it is one of the most valuable.

Regarding MM and AGI, this is the MM blog not the AGI blog so I'll be brief. MM goes nine tenths of the way to solving the AGI problem, though perhaps not the "safe" AGI problem. As far as "safe" goes this is my opinion: you are not making a new kind of general purpose problem solving device, you are creating a new species of intelligence far beyond humanity. I don't think the coming superhuman AGI is going to need our advice on how it should behave or what it should think. Your whole presumption that we should build in our needs as paramount to its, is frankly, presumptuous. Evolution doesn't work that way.

Mike Treder, CRN

Evolution doesn't work that way.

Correct. But evolution by natural selection is frightfully destructive, culling 99.9% of all species in favor of the few who can survive to reproduce.

Are we willing to risk those odds? Is there an alternative? Can we take control of our own evolution, including the selection of our (benevolent) successors?

My answers are: No, Maybe not, I hope so!

michael vassar

Correct on all three points Mike T
Mike D: AGI can be any of a number of things, but it's trajectory in the transhuman has little to do with natural selection. I won't try to convince you about AGI, as I said, just focus on nanomedicine, but no, I really do what MNT enables. Thing is, I also know a bit about what existing technology enables, and about what humans are actually capable of consuming. Modern roboticized factories add many hundreds of dollars of value to the components they process with each hour of labor consumed. Basically, scarcity of the sorts that primitive MNT will eliminate is always artificial anyway in the modern world, except arguably with respect to energy scarcity.

Phillip Huggan

The main advantage MNT has over AGI is that it is almost impossible to accidentally use MNT in a risky manner. Whereas the default AGI scenario seems to be very scary.
The main disadvantage MNT has that AGI doesn't share, is that the intentional misuse of MNT is much more likely than is the intentional misuse of AGI. This makes MNT political in a way that AGI isn't.

SIAI is working on a way to channel an AGI's actions in an odedient manner. This is certainly a prerequisite, but hardly sufficient if the FAI goal-structure that forms the boundaries of AGI actions, channels to an extinction scenario.
I think there is a great deal of singulatarian confusion regarding what is good. Minimizing extinction scenarios (and shrieks) is a very different goal than is attaining god-like powers; they are being dangerously melted together in H+ circles. The Oracle/PAI AGI model seems by far the safest to me, but there aren't any researchers advancing it. I didn't even learn of it until a novice newbie posted on sl4 a few weeks back.
That being said, leading AGI proponents have neglected surface free-energy considerations in diamondoid formation models. I didn't learn of this until doing my own diamond research. And CRNano appears to be the only organization that even realizes there needs to be MM administrative structure models and MM implementation documents drafted. Feymann's lecture in 1959, Drexler's book in 1982. No perception of risks until Gubrud's 1997 essay and CRN's 2003 inception?
There is much to be done regardless of whether you are in the AGI or MNT camp.

Phillip Huggan

Last paragraph should say: leading MNT proponents, not AGI.

The comments to this entry are closed.