February 27, 2006

Economics: Ecological Economics

WorldChanging: Joshua Farley, Ecological Economist, an interview. Not as in-depth as I was hoping, but it's still quite interesting. He talks about how he introduces students to the concepts (by immersion in real problems) and about sustainable economic metrics to replace the GDP. We need to see a lot more of this kind of stuff in the future. I am sorry that I missed the recent Santa Cruz Future Salon where they talked about a "agricultural" approach to energy (rather than our current hunter/gatherer system).

February 27, 2006

Transhumanism: What Would Radical Longevity Mean? Redux

WorldChanging: Another World Is Here: What Would Radical Longevity Mean?. I blogged about this very essay about a year and half ago, and it's interesting to revisit both the article and what I said about it at the time.

First, it's obvious in hind-sight that the essay had far more depth than I commented on at the time - the pill popping was not the focus of the essay, and indeed Jamais said in the article he thought it was the most unlikely of the four scenarios. Since then of course, drugs targeting that very gene have been developed and are being tested...

But the larger question of how we are going to pursue longevity and what it might mean to us and our society are much more interesting to me now.

The "Holy Fire" scenario is indeed the most likely route to a vastly increased longevity - it's the route advocated by the SENS strategy and the MPrize.org reversal fund.

As Jamais said, that type of technology is likely to lead to "rebirth" type scenarios, which I think makes for an interesting world. Youthful looks may no longer mean inexperience - and conversely aged individuals are more likely to be poor and lower class (since those with more money become younger looking!). Also, his point about the social implications approaching slowly holds in this scenario as well: it will take decades before there is any substantial population of "young" elders. This is one of the reasons that I am not terribly worried about how society will adapt - we will adapt slowly over time, just as we did during the longevity boom of the 20th century!

The question of the cost of the treatments is interesting. In Holy Fire, they require everything that a person has - at least in the initial stages. The first 100 people for each treatment bet not only their lives but also their fortunes - and the world is a far more conservative place, capital wise, because of this. But can we really expect such treatments to cost multiple millions of dollars?

I'm no expert, but I think that medical costs are likely to go *down* not up as we enter the biotech age. Taking people apart with scalpels and giving them narrowly targeted drugs which have nasty side effects is expensive. Controlling our (cellular) health directly via gene therapy, nano-tech robots, and stem-cells is likely to be considerably cheaper: all the real work is being done by the technology (whose price always falls rapidly on introduction to the mass market), rather than by the highly trained medical staff working in expensive facilities. The biggest hitch to that vision is how diagnosis is preformed - which treatment is to be used? Eventually you can automate that with technology as well, but I see that as being much further in the future...

Socially, the world of extreme longevity will be an interesting one. I think there will be a substantial chunk of the population that will choose to die natural deaths - perhaps as high as 25% in the ultra-religious USA, but probably not lower than 5% worldwide. Suicide and euthanasia will have to become socially acceptable, first because they are likely to become far more common as a cause of death, but secondarily because there are good reasons for needing them: some things which people now tolerate "a lifetime" of will become unthinkable to tolerate for an eternity...

And that I think brings the most interesting aspect of super longevity to light. When you know you've got (potentially) forever, how does that change your actions now? Do you save more? Do you plan more? Do you act less rashly (i.e. not marrying quickly?). Are you more concerned for the environment? Given a vision of living for thousands of years, does the transitory nature of physical objects (even big ones like houses!) make you less attached to possessions? Do you try to improve yourself more diligently? What will your future self 100 years from now think about your current hobbies? Will there be regret that you wasted so much time watching TV? Or will time wasting become the name of the game, because you have so much of it? If you really set yourself to it, what could you achieve in 1000 years?!?

How about your attitudes to government? If you can expect to be alive 100 years from now paying the national debt, do you pillory the politicians who mortgage YOUR future? Historically any particular nation/government/regime/dynasty lasts only a few hundred years - do you become more tolerant because you expect to outlive the problem? Once you start thinking about it, the long-term stuff really changes how you act and what you believe.

And of course, what about all those things which require people to die or move on in order for progress to occur? Thomas Kuhn made this famous in science - the paradigm can't really change until the oldest generation (who don't get it) die, and the younger generations replace them. It isn't so much an issue of entrenched power, but rather of entrenched ideas. Our brains are big complicated neural networks and despite the amazing plasticity in some cases, the truth is that much becomes set in stone after a few decades. Training a brain takes a lot of time and energy - and retraining it out of old, bad, patterns takes much longer. Or maybe we will have treatments for that too?!?

Anyway, I look forward to a future in which more than just transhumanists are contemplating these issues and taking seriously the idea that vast longevity could be theirs. We've come a long way in the last five years, and I think the next 5 will see super longevity really impact on the mainstream.

February 27, 2006

Sustainability: Driving the Solution

Driving the Solution: The Plug-In Hybrid Vehicle (warning: PDF). Makes the case for plug-in hybrids - I especially like the chart on the next-to-last page showing that the impact on the electricity grid is largely at night, when utilization of generation capacity is very low anyway. What we need is for a large utility to team up with a large auto-maker to get the ball rolling here...

February 25, 2006

Awesome Events: Strong AI Future Salon

Future Salon: Strong AI with Eliezer Yudkowsky. The best future salon in recent memory! The topic was vast and abstract, but Eli did a great job making it clear and easy to follow. I think he's had considerable time to polish that presentation - on audiences with less familiarity than ours as well. A+ dude! Read on for the interesting ideas that I took from the presentation.

First I want to state that there were hundreds of great ideas and that he covered a lot of ground that other people might find fascinating - such as evolutionary psychology, the singularity, the nature of human intelligence, friendly AI, and innumerable interesting analogies that make grasping these ideas far easier than it would otherwise be. I am not going to cover much if any of that here - you can go read his stuff or look it up on the interwebs yourself.

Instead I want to focus on several of the side issues (or supporting lines of argument) that Eli had to make in order to get to his understanding of friendly AI.

The first is his definition and measurement system for intelligence. He views it as an optimization process - a thing which can "hit a small target in a large search space". Example: of all the ways to organize the atoms in a Corolla (~10^30), the Corolla is obviously a pretty good solution the problem of how to get around. The intelligence embodied in that solution can be "measured" by taking the negative natural logarithm of the size of the target (organization of atoms which function as transportation devices) over the size of the possible search space: all the different ways to organize 10^30 atoms. Obviously in this example those sizes are hard to estimate, but in simpler systems you can get some interesting answers.

For instance, Eli estimated the "intelligence" embodied in the most primitive life form by looking at it's probably of formation. If you believe that the odds for the formation of that replicator from random organic molecules are 1 in a quadrillion (1e-15 = ~1/2^42) then you're saying there are 42 bits of "optimization process" (intelligence) required to produce that object. Eli believes that "emergence" can explain at most 100 bits of optimization process - anything more requires evolution or intelligence itself... This metric becomes important later to understand how much intelligence humans possess.

The next fascinating analysis he undertook was to show the limitations of evolution. How fast can beneficial mutations spread through a population? What is the rate of new mutations? Eli claims (based on a paper that he referenced) that evolution can consolidate at most 1 bit of information per generation in a species. Since a base-pair requires 2 bits, clearly evolution is quite slow to make changes :-).

An interesting corollary is that there is an upper bound on the amount of information that evolution can maintain against the degenerative effects of mutations... and this limit is much smaller than you might think, at about 100 million bits. At 100M bits, no positive evolution could occur, because the entire 1bit/generation selective consolidation of information would be required to just maintain the status quo. Therefore the effective upper limits for any real species is considerably lower. It so happens that the human genome contains information equivalent to a few tens of millions of bits - we are likely at the complexity limit of evolution for DNA!

This implies that for every bit of complexity that we gain in e.g. the brain, we must lose somewhere else, e.g. in muscle strength, etc. This limit is of course controlled by things such as the mutation rate - but those constants are themselves trade-offs (in that less mutation in general would mean slower evolution - so supporting more complexity would come at the cost of taking longer to actually use that complexity productively!).

So to sum up he showed this slide:

Emergence
is enormously slower than
Evolution
is enormously slower than
Intelligence

and elaborated that not only are they slower but they have nasty upper bounds on the complexity they can create. As I mentioned, Emergence is about 100 bits, Evolution about 100M bits, and although he didn't state it, I expect that human-level intelligence probably maxes out as well, perhaps at 100T bits? The "optimization process" ability's are similarly ranked, with Emergence at millionths of a bit/year, evolution at 1 bit / generation, and human intelligence at roughly 1 bit/second.

There is a *forth* entry in that series, after intelligence, and we call it AI: Mind designed by Mind rather than by evolution. I believe that AI will have a better name for itself than we do on this series - i.e. that AI will be as different from Intelligence as Intelligence is from Evolution. "If you think we're the product of intelligent design, you ain't seen nothing yet!"

Finally, Eli had a aside about "generalizing over all possible minds", to warn against it. Anthropomorphism (thinking they are all like humans - bug eyed monsters will NOT be interested in human females) is of course the big danger, but even thinking they will all be the same is NOT VALID. AIs will differ profoundly from type to type - Bipping AIs will not be the same as Glooping AIs will not be the same as Freeping AIs. So saying anything in general about AIs is a sign you haven't thought about the space very well.

And, of course, Eli got to talking about "Friendly AIs", the "small target" of possible minds that we would like to find in the "large space" of all possible AIs. He has some interesting thoughts about how such a mind needs to be structured - including the mathematical problems relating to how the beast evaluates possible changes to it's own source code (including the special infinite-recursion case of modifying the code which modifies code...). His thinking here is clearly still under development - and he made the expected pitch for money and volunteers to help him out.

To get back to intelligence as an optimizing process: even though an optimizing process may make the immediate steps (moves in chess) become less predictable, that same process makes the outcome become MORE predictable (winning the game) if you understand the optimization target. Therefore:

"A smarter-than-human entity whose actions were surprisingly helpful could produce a surprisingly pleasant future. We could even assign a surprisingly high probability to this fact about the outcome."

He address several critics and their ideas, including a memorable "fallacy" approach to the "but a friendly AI will just change itself into an unfriendly AI, so why bother designing for friendliness?". Basically the Giant Cheesecake fallacy is thinking that just because an AI *could* do something (convert all matter on earth into a giant cheesecake) it *will* do so. It will NOT do so unless it has some motivation for that action. And motivation is one of the design inputs to an AI - just as human biology has strong impacts on our motivations (think sex, etc), AI designers will have strong inputs to the AI's motivation. And when you want something, you will redesign your future selves to keep wanting it - otherwise you know you won't achieve it and ipso-facto don't want it now. Motivations are quite stable under self-modification - although not perfectly so in humans. Eli believes that they can be made perfectly stable in AIs - although the "proof" for this belief has not been found.

And, an interesting AI related phrasing of Moore's Law: every 18 months the IQ required to make AI drops by 1 point. We have the hardware to do it NOW, but nobody understands mind anywhere near well enough to actually implement AI yet. Eli hopes that educated teams will solve the problem in the right way long before dumb teams solve it wrong way just by throwing money and computational power at the problem...

February 23, 2006

Sustainability: Defining Sustainability

Green Car Congress: The Slippery Task of Defining Sustainability. Excellent essay on the roots and definition of sustainability. Also talks about "The Natural Step", and it's four system conditions for a sustainable world.

February 23, 2006

Health: Anti-Aging via CR/SIRT1 drug

Caloric Restriction Calculator - handy. But, I do wonder why activity level should impact how CR'ed you are - is this a result that has been tested in e.g. mice? If it's true that just increasing your activity without increasing your caloric intake makes you CR'ed, then CR is clearly NOT just a "metabolic usage causes damage" effect. This is supported by this Scientific America: Unlocking the Secrets of Longevity Genes article, which details how a gene named SIRT1 (and it's analogs) may be a master regulator for the rate of aging in most lifeforms on earth. Fascinating stuff...

The scientific american article says that:

<<
Nevertheless, those of us already alive could live to see medications that modulate the activity of Sirtuin enzymes employed to treat specific conditions such as Alzheimer's, cancer, diabetes and heart disease. In fact, several such drugs have begun clinical trials for treatment of diabetes, herpes and neurodegenerative diseases.
>>

These drugs may have the "side effect" of modulating the rate of aging in people who take them. I wonder how much money 30 extra years of life is worth on the black market?

February 23, 2006

Transhumanism: Human Chimera

Damn Interesting: The Not-So-Legendary Chimera, about a woman whose babies DNA tested as not hers... turns out she's a Chimera and the babies came from her *other* set of DNA. I wonder if anyone has thought to study such Chimera's for insight into how their immune system copes with the unusual situation.

Note: it doesn't say in this story, but the way they knew the babies hadn't been switched is because their DNA "mother" was a "sister" of the mother: except that she has no sister. The same would have been true about the comment at the end of the story: if the man had been a chimera, the baby would have tested as being of the *brother* of the husband, and it would have been verified either immediately (no brothers) or quickly (not actual matches) that the husband actually was the father.

More interesting information (especially about cats - great examples of Mosaicism).

February 22, 2006

Sustainability: Melting Greenland Glaciers?

TCS Daily: Ice Storm. You've probably seen all the recent news about how the Greenland Glaciers are calving and melting at a rate 3x as fast as even 20 years ago. Yet as this article points out, recent research has also shown that it's snowing more in Greenland and the height of the glaciers is increasing, not decreasing... combine this with some historical data showing cycles in weather with a period of a few decades, and you have enough data to see that the alarm the recent Greenland news has triggered is probably misplaced. By no means do I think that global warming isn't an issue - but we need to be careful of the things we point to as evidence that is it occurring.

February 22, 2006

Economics: Asking Questions

I Will Teach You To Be Rich: Smart People Ask Questions, which describes in detail a sin I am guilty of: not asking for help, and not asking questions. Sounds like it would make a good life-style change :-)

February 22, 2006

Sustainability: Diesel Hybrid's coming

Technology Review: The Next Prius? which describes how PSA is bringing diesel hybrid's to showrooms starting 2010. Why not next year? Anyway, at least some action is starting to occur on this obvious sustainability solution.

February 20, 2006

Transhumanism: Age Lab

Sacramento Bee: Laboratories' age research is growing up, a look into the world of anti-aging research. There is a surprising amount of money being spent even now - this one facility has a $25.6 million budget. It would be an awesome place to work!

February 20, 2006

Sustainability: Al Gore's "An Inconvenient Truth"

The Observer: Al Gore: the second coming. Apparently Al Gore has a new film out about global warming called "An Inconvenient Truth", and it's better than good: it's great. It was the talk of Sundance, as was Gore himself. Think how different the world would be if he'd been elected president in 2000! At any rate, I will certainly be going to see this when it opens at the local independent theatre.

February 18, 2006

Transhumanism: Demos: Better Humans?

Demos: Better Humans? The politics of enhancement and life extension. See their Press Release which is a pretty good summary by itself.

<<
A public debate is needed now about the potential for new technologies to make us ‘better than human’ according to a report published today by Demos and the Wellcome Trust. Better Humans? argues that policy makers and the public must address the consequences of technologies to enhance the human mind and body, including memory-enhancing drugs, genetic selection of children, and dramatic increases in life expectancy.
>>

There are 13 small essay's on the topic, ranging from an intro to transhumanism, an interview with de Grey, to presentations by transhumanists. Overall I'd say the presentation is pretty balanced. However the quality of the material is mixed - some are clearly far better than other.

I like "Better By Design", which shows that in fact many of the concerns about policy lagging tech are vastly overblown - the UKs experience with PGD shows that in fact legislative actions can be well ahead of any possible issues - PGD is still not widely used, but the policies to head off all manner of bad outcomes have been in place for over a decade.

I also like "Nip/Tuck Nation", for it's clearly presented tale of "enhancement" technology moving from scorned outcast to mainstream consumerism. I expect that many other enhancement technologies will follow a similar path, for similar reasons. Ours is a personal improvement culture, and I think it's even possible to argue (as Kurzweil does) that ours is a personal improvement *species*. The same logic which makes us pay attention to our health will make us pay attention to possible improvements above and beyond what nature can offer - and the slope is not just slippery it's enticing!

I like "Does Smarter Mean Happier", although the first half (about the relationship between IQ and suicide) is more interesting than the last half (about "feminine" virtues).

The introductory essay "Stronger, Longer, Smarter, Faster" is also quite good: it gives a good overview of the coming technologies, while also showing the myopia that mere extrapolation and generalization induce in transhumans. We are by-and-large highly educated male white technology workers - and that is both our strength (who controls the levers of power?) and our weakness (we fail to see the world as it is; instead seeing it as we experience it from our positions of privilege).

The most often discussed topic in the essay's is education - think of the children! In parents race to give their kids an advantage, many things are bring tried, many policies are being implemented, many drugs are being prescribed. Children have always represented the future (they do become it), and as such focusing on them makes sense when talking about Better Humans.

Yet I would have liked to see more coverage of the possibilities for longevity - in many ways the thinkers have failed to address the idea that the present powerful may never give up the reigns of power. For instance, imagine 30 years from there is a figure equivalent to Greenspan in an important appointed position. Further imagine that longevity science has achieved "escape velocity" and that this individual (along with millions of others) will not age or die. Will he hold the position forever? What could make him leave it? Will generations coming of age later than him resent his "permanent" hold on power? I think the clear answer is to make all positions have term limits - or at least periodic reelections - yet this works only in public institutions. What about corporate positions?

At any rate, the policy challenges facing a Better Humans future are many and numerous. The UK is handling the challenge much better than other western countries - many of which, like the USA, seem to be completely ignoring the problem or even retrogressing their level of governance with anti-future policies. As with PGD, we can at hope that at least we will be able to use the UKs good governance on this front to get a leg up when these things become more important.

February 17, 2006

Transhumanism: ImmInst: Exploring Life Extension Movie

ImmInst.org ~ Film Project, a great film interview about 60 of the leading lights in the transhumanist community. They address all the big issues in ~5 minute chunks. I think it would be a pretty good primer for someone who didn't know much about transhumanism and the quest for extended lifespans - it even includes intros to nanotech, nanomedicine, cryonics, and the general science behind fighting aging. All I can say is BRAVO!

February 16, 2006

Future Studies: Future Transport

Intelligent Infrastructure Futures via Soft Machines.

They put forward 4 scenarios for the future:

Perpetual Motion
Urban Colonies
Tribal Trading
Good Intentions

Good reading. I prefer some combination of the first two, but fear that "good intentions" may be what we achieve...

February 16, 2006

Technology: Apple's Video iPod

I, Cringely: Apple Saves Blockbuster via slashdot. Cringely speculates that Apple will install video transfer stations at your local Blockbuster, as a way to make broad-band free households want the rumored Video iPod. It's not a bad idea, but I can't see that it would work: those types of people are unlikely to buy an iPod, partly because they won't be able to download songs either, but more profoundly because they are *late adopters* in a tech cycle and iPods are still early majority. A slashdot commenter points out that Apple could achieve the same objective by installing kiosk's in various places, sort of like pop machines only they would sell iTMS material. Given the equivalence of that scenario, I think you can see the foolishness of Cringley's idea...

All of this of course assumes that Apple does in fact come out with the rumored Video iPod. I think a video iPod is highly likely, and that the proposed form-factor is likely true. The wide-screen aspect would allow iTMS to sell the required two versions of each video in a consumer intelligible manner: regular "iPod format" and "iPod wide-screen". The major challenge for Apple will be making the click-wheel function properly on a touchscreen: haptic feedback is necessary! Nevertheless of all the Apple rumours that I've seen the Video iPod with touchscreen one strikes me as highly likely to be true.

February 13, 2006

Sustainability: SkySail Trial

Treehugger: Beluga Group Signs Contract for Sky Sail Power. Powering the world's largest ships with the world's largest kites! For fun and profit! I believe that harnessing wind power on the open seas will eventually return to favor - the question is only how expensive fuel has to become... Kites offer an ingenious way to accomplish that with minimal structure on the decks.

February 13, 2006

Sustainability: Ultracapacitors

TerraDaily: MIT Researchers Fired Up About Battery Alternative. A capacitor is basically a device which accumulates electrons and releases them at a later date. They are generally used as filtering devices on printed circuit boards (those green things inside your computer), because they "average" a flow of electrons, taking when there are too many and releasing when there are too few. An ultracapacitor is like a capacitor, only on steroids. Generally the idea is to store *so many* electrons that the device becomes like a battery. However, the limitation has always been power density and cost: it's hard to cram electrons tightly into any device, let alone a cheap one.

The MIT researchers are all fired up because apparently carbon nanotubes show potential as a ultracapacitor! What are those little suckers not capable of! If their research pans out, they claim that ultracapacitors could be as energy dense as the best-of-breed batteries today, but offer the ability to discharge far more rapidly. The article even claims that they have high efficiencies in charging and discharging - surprising, but I guess on reflection packing electrons together is likely to be fairly reversible and thus isentropic... Anyway, here's hoping that the research eventually pans out, since the world would be a lot more sustainable with better energy storage devices that don't require heavy or toxic materials.

February 13, 2006

Health: Resveratrol - amazing, but not ready yet

New Scientist News: Red wine's anti-ageing ingredient does it again. via Fight Aging Resveratrol has been shown to actually delay ageing in cell cultures, and now even in live animals (fish). It's effects appear to be similar to that which CR achieves (potentially even via the same mechanism related to the SIR/SIRT1 gene).

However, studies in humans show that the substance is almost entirely destroyed in our digestive systems. Furthermore, the amounts required to get the big effects on lifespan are hundreds of times that found in red wine - as the link says: the highest dose of resveratrol given to the fish equated to a person drinking 72 bottles of red wine per day" and that's assuming your digestive system doesn't destroy it! Consuming the resveratrol in concentrated pill form might make that task more manageable, but resveratrol pills are expensive, certainly far out of my supplement budget. Kurzweil claims to use 400mg/day, this amount would cost at least $50/month - and I don't believe it will have any effect, given the quantities the studies I have seen claim is necessary.

Nevertheless, I believe that eventually there will be a form of resveratrol (or a cousin) which actually works in humans, i.e. there will literally be a pill which can extend your lifespan by 30% (alternate phrasing: can slow the rate of aging by 30%). The effects that have already been shown for resveratrol PROVE that there are fairly simple high-level genetic effects which determine at least a substantial portion of how fast we age. Modifying those genetic effects via drugs will be *easy* with technology that we will possess in a decade or two. For instance, the same sort of medgadget that could dispense insulin to diabetics could be easily adapted to dispense resveratrol++ to transhumanists. And there are millions of diabetics who are driving the research on that medgadget...

What will be the implications of a medical gadget which dispenses a drug that can give you an extra 25 years of life?

February 13, 2006

Transhumanism: Positive Review of Transhumanism

FT.com: Who would not seize the chance to live to be 150?, which is more evidence that transhumanism is going mainstream at a fast pace. Best quotation (emphasis mine):

<<
Provided enhancement technologies are carefully regulated, and opened up to genuine public debate, there is **no reason** why they should not enjoy widespread public support.
>>

February 8, 2006

Health: CR versus Exercise

A few weeks ago I blogged a question about the relative benefits of exercise versus caloric restriction. Now here this: April's CR Diary: Exercise Does Not Slow The Aging Process. Which makes a fair amount of sense - if it did, we'd hear about all those Olympic centenarians! April strikes a balance far to the CR side, obviously... 1050 calories/day indeed!

February 8, 2006

Politics: Iconoclasts and Cartoons

TCS Daily: The New Iconoclasts where the author compares idol smashing in Christian history to the recent Islamic outcry against the Danish cartoons. The conclusion, worth repeating here:

"In short, the new iconoclasts are winning -- they are realizing that they have the power to make us suppress any image that they find disagreeable to their stern and mirthless fanaticism -- even if it is just a funny cartoon in a paper published in a cold corner of Europe, far far from Mecca.

Either Muslims need to begin to get a sense of humor, or we need to became a great deal more serious. "

February 8, 2006

Politics: Bush's Energy Promises

WattHead: Bush Highlights Energy in State of the Union Address - Admits "America is addicted to oil", but Bush's follow-up budget proposal is deeply flawed from any reasonable "green" viewpoint. There is far more money for the "hydrogen future", "clean coal" and "nuclear power" than anything else. In short, Bush is giving more money away to large transportation and energy companies, rather than focusing on actually solving the problem by reducing consumption (energy efficiency, better CAFE standards) or deploying/supporting technologies which work right now (like wind power, hybrid cars).

February 6, 2006

Economics: The Global Toothbrush

Spiegel: The Global Toothbrush a tale of "forty-five-hundred employees, 10 countries, five time zones". If you want to see what globalization is really about, this story is an on-the-ground look with pretty good details.

February 6, 2006

Sustainability: Green Roofs in Toronto

City of Toronto: Green roofs and news release of same - go Toronto!

February 5, 2006

Sustainability: Industrial Organics

Rabble: Does Corporate Organic Change Organic Landscape? via sustainablog. So last month I argued in favour of Whole Foods Markets (WFMI), and one of the arguments was that big food stores like WalMart and Safeway would not be able to source "Organic" products for their shelves - it was a matter of supply and demand (supply being woefully insufficient). But apparently I was wrong. Industrial agriculture has figured out how to grow hundreds of acres of broccoli organically - as well as whole grains, and soon practically anything. They harvest it mechanically, then ship it hundreds or thousands of miles to large buyers - who apparently include WFMI itself. Is the end near for the "meaning" of the term organic? Will WFMI respond to these concerns?

As the story said, on the one hand this trend represents success for "organic": it is going mainstream and cultivated acres is growing exponentially. But on the other hand, "organic" is selling out, and becoming less about healthy food and sustainable agriculture and more about branding, which is the mere appearance of value. Consumers buying "organic" labelled goods at WalMart probably don't even know what the term really means: but they get a positive, healthy vibe from it and thus might pay a bit more for the label (or choose an equivalent organic product over a non-organic one).

I believe that ultimately this is a positive trend, in that while "Industrial Organic" may not bring everything we'd like to the table, it will be a net improvement over the status-quo. The agricultural system of the developed world will move closer to a sustainable basis.

For WFMI, the mainstreaming of organic is both an opportunity and a challenge. It's an opportunity because it means the number of consumers interested in their "niche" is exploding. It's a challenge because unless they can deliver the growth to capture those consumers, they will loose them to WalMart's organic line, or Safeway's organic line, etc. Ultimately the product difference between WFMI and Safeway is eroding, and that could spell doom for the company.

However, they could respond in a multitude of ways - the most obvious of which is to emphasize the other traits of "organic" that are being lost in the transition to mainstream. Localization is one. Actually healthy food (as opposed to a brand connotating healthy) is likely another. And on this front I expect they are already quite active. There will likely be a successor to the term "organic" which includes what organic already mandates, and also numerous other criteria.

Never-the-less my optimism about the company has been somewhat tempered by this story, as their competitive edge is less secure than I had thought. But I don't believe that the impact will be near term - this looks to be a trend that will take at least 5 years and likely a decade to play out, meaning that there is considerable upside to be had in WFMI yet before this even has to be faced as an issue - and as above, I believe that there are numerous ways they can face it and survive to thrive. So I am holding the stock.

February 5, 2006

Health: Flu Vaccination in America

TCS Daily: We Have It Coming, an interesting editorial explaining how it is that American's have to import practically all of their flu vaccine shots (reason: politics fucked up the domestic suppliers). As you know, I am in favour of government meddling in health care - I think the evidence speaks for itself when you compare countries with socialized health care to countries without (i.e. America). I think America's problem is that they are trying to have it both ways: not controlling the important things (like making sure people actually have access to affordable health care!), while over controlling unimportant things, like the *trivial* cost of vaccines. This is a sign of bad governance, which is rampant in America.

February 4, 2006

Future Studies: Computers in 2020

I've been watching the computer industry my entire life - my parents bought me a Commodore 64 when I was young enough that I don't remember any childhood before the computer. (I was maybe 8 when we got it?). Later we upgraded to a PC, and I've "upgraded" my PC at least 4 times since then, most recently to an iMac - each time was a significant upgrade, because I usually wait a few years each time. Several times in fact the new machine was almost 10x more powerful than the last, with 10x as much storage space.

Obviously this trend will continue. In fact the determinism of improvements to computational systems is legendary: Moore's Law. And yet, it's most recent turn surprised a lot of people: instead of getting faster in cycles per second (frequency) as had been happening for years, the last three have seen frequency increases fall by the way side while instead multiple-core processors have become the trend. What does this portend for the future and where can we expect to be by 2020?

1) Frequency will increase, but at a much slower pace than before. Frequency is a primary variable in how much power a processor dissipates, and that more than anything is why the roadmap is now multi-core. I believe we're likely to be operating in the 4 to 6GHz range in 2020. More importantly, we're likely to be operating at <20 watts even though the capabilities of the chip are vastly greater: cold enough to be cooled passively.

2) There is a limit to the usefulness of multiple-cores, especially on the desktop. I believe that Intel and AMD would be foolish to EVER place more than 4 cores on a desktop or laptop processor chip - additional cores would not have ANY value to the user of the machine in terms of actual software speedups, etc, because users (humans) rarely do more than one thing at once. Even quad-core may be too many. Obviously on a server the situation is different: but that means Sun, not Intel...

3) Therefore, because transistor count will keep increasing according to Moore's Law and the ever-smaller technology nodes (65nm now - 45 nm is already planned, not to mention nanotech, etc...), there are going to be transistors to do things with. There is already a lot of buzz about "system-on-a-chip" SoC stuff in the embedded world, and that will extend out to general computing soon. AMD has already integrated the "North Bridge" memory controller into their processor dies and achieved impressive gains in terms of access-to-memory performance. This paves the way for more improvements of a similar type: every time you get more transistors, just integrate another device currently on the motherboard or other daughter boards, in order to improve overall system performance. This parallels the trend whereby the motherboard has been subsuming tasks previously requiring daughter boards.

The first obvious target is the main memory. Memory has consistently been using about 2x as many transistors as the processor, so it could take several years of progress after quad-cores before the L1, L2 and L3 caches grow big enough to displace the main memory completely.

Then there is the graphics card and processor and memory. I believe that there will long be a role for independent gaming graphics processors, but the time is near when office-type graphics requirements (basic 2D and 3D graphics acceleration) could easily be placed on the CPU die.

Flash storage. It is widely perceived that it will eventually come to replace HDs, especially in mobile applications like laptops. Flash storage is more reliable, uses less power, takes less space, and has better performance than HDs. I can see in five years that there will be large flash chips in most laptops - say 20GB or more - and no HD. Users wishing more storage can still have a regular HD, probably external (think iPod like - it will have functionality on it's own, like multimedia stuff, since that will likely be what it stores most often). But eventually that flash chip too will be subsumed by the CPU die in order to increase performance and reduce size and cost. I think there is a "low" limit to how much flash you might integrate into the CPU die by 2020 (20GB?), but as with the graphics stuff, users with exceptional needs can always add an external flash chip with way more capacity. The point is to get the OS into that 20GB, in order to get the performance edge that proximity will give you. Storing your vast multimedia collection off-die won't make any performance difference and therefore on-die flash is unlikely to grow much beyond what the OS + minimal user software and storage require.

Miscellaneous stuff - network controllers, USB hubs, battery management stuff, etc - all of this slowly gets taken in as well. The wireless chip may have to remain independent if only because it's RF emissions need to be some distance from the other stuff to avoid interference. It will probably be "software controlled radio" wireless - but that's another essay.

And so by 2020, the "office"-type laptop may be reduced to a single "big" IC, the quad-CPU-GPU-Memory-Flash chip sold by Intel. Combine with awesome OLED display, batteries 2x as energy dense as anything available today, and a gigabit wireless chip, and you can imagine that the "laptops" will be pretty darn amazing. I envision them being like the "Star Trek" tablets, wirelessly connecting to keyboards & mice when in range, touchscreen control otherwise, have battery life measured in DAYS, and compute circles around anything sold today. All for less than we pay today. It's also note worthy that Intel and AMD will be taking ever greater percentages of the PC revenue, as they supply more and more of what goes into a computer. That's the vision.

What could make the vision not happen according to plan?

A) The rise of thin-client systems. With great wireless networking, it becomes more and more compelling to just do all the processing for an office at some central location and make all the user terminals "dumb", i.e. lacking everything but screen and networking (and therefore cheap). I believe this will not come to pass because users will value mobility too much - and lag & bandwidth become a serious issue if you're 10,000 miles from your CPU. But thin-client could still take a substantial share of the market, especially for "desktop" systems (which are on the decline anyway).

B) Greedy multi-core killer applications. If "office" software which can actually use 8+ core systems develops, then Intel might rationally choose to keep placing more cores rather than subsume other components of the computer. But what software could that possibly be? I believe that AI is the only serious potential on that front - and AI is an entirely other essay.

C) OS requirements become absurd. If the requirements of the operating system grow extremely fast, it may not be possible to fit enough memory and storage unto the CPU die by 2020. To be honest I think this is unlikely - software requirements have begun to lag hardware in recent years. But perhaps Vista will show that MS is hopelessly inept and has lagged & bloated the OS beyond all reason. I wouldn't put it past them. Ultimately this just delays the vision rather than changing it - instead of 2020 for the single chip systems, we might be looking at 2030.

D) Moore's Law fizzles out. There is no successor to lithography, and we're stuck at say a 35nm node and 10 billion transistors per IC. Highly unlikely, given what we know about accelerating change and the way technological paradigms cascade, but obviously if it did happen this vision would fail because an individual chip couldn't possibly subsume enough functionality.

E) New types of necessary and greedy processing tasks are invented. For instance, was it foreseeable in 1990 that graphics cards would require so many transistors in 2005? It could be that there is some new thing arriving "tomorrow" which will require large amounts of processing, memory, etc and that would therefore represent a barrier to this single-chip vision. For instance, how about neural interfacing? As with (C), this would merely delay the vision, but depending on the task it might be a considerable delay...

What does the vision tell you about the computer industry?

A) You should eventually (2012?) sell companies like HP & Dell, who make their money by doing all the "integration" work a computer currently requires. This task will become (and is becoming) easier and easier, and hence will command lower and lower margins as time goes on. Eventually their task becomes so easy that even Grandma can assemble her own computer by buying a chip from Intel (it comes with a very simple motherboard and is preloaded with OS and apps), and a display from LG (it comes with built-in battery and nice case). Insert monitor cable in connector A, slide PCB into plastic case. Presto, computer. Why pay Dell 5%? Dell's market will retreat to higher-end computers still requiring external chips and devices: graphics workstations, servers, media storage computers, etc. But this will be a small and shrinking fraction of the total computer market, as the single-chip solutions move steadily upmarket. And, Dell and HP would share this market with IBM and Sun (see E).

B) You should buy Intel and AMD, for the inverse of the reasons in A. Also buy companies which make OLED displays, touchscreens or batteries - although sadly they tend to be conglomerates, so it's not as obvious a power-play.

C) Innovators in technology already see the early stages of this vision and are working on the technology needed to make it happen. Apple is a perfect case in point. There is a lot of money to be made in leading the pack on this vision, and I expect to see Apple profit from that. And sadly, Microsoft will probably profit as well, short of some kind of revolution in the computer industry... Eventually, of course, even Apple computers will fall to the problems enumerated in (A), but by then Apple may be a very different company. Can you see Dell or HP pulling off a similar transformation? I can't...

D) Google, Yahoo and other internet information providers will be largely unaffected by this, although of course they will continue to benefit from increasing network speeds and deeper penetration of computer usage into society. As such they make good investments, if you can catch them when their valuation is low...

E) IBM and Sun have an opportunity to drive the thin-client market (maybe now is the time?), and to profit from data-center operations. All those wireless tablets may not be "thin" in the usual sense, but they will share a common need to access external resources like vast amounts of storage space. Consumers are unlikely to want to run a "server" in their house, but I bet lots and lots of them would pay an extra $10 on their internet bills to have a terabyte of storage they can stream to themselves on demand. However I do not believe this can save them. I would sell both - for the same sorts of reasons as in (A), only at higher levels. Computers are becoming more resilient and less reliant on highly-qualified personnel. Taking care of the high-end systems is an ever-less-difficult ever-less-common job.

February 1, 2006

Politics: Since Sliced Bread

SinceSlicedBread.com: what's your common sense idea? Via Wired News. It was a contest where regular people entered a short policy idea, and a community of experts and interested users determined winners. Since 22,000 ideas were entered, the selection process must have been very difficult! Never-the-less 3 winning ideas were chosen and $200,000 was given away. As a first effort I think they did quite well, but there are some obvious improvements to be made...

First, the quality of the winning ideas is still not terribly good. I think this results because there is no way to collaboratively edit entries and change them in response to feedback, etc. In short the ideas remain trapped in their original expression - which as anyone who writes knows can often be quite bad. So, the next version is in bad need of collaboration tools, editing tools, etc. I think perhaps wiki software could be adapted.

Second, it's clear that almost all of the 22,000 ideas are redundant with each other, i.e. there are actually only a few hundred *different* ideas, yet because of the way the contest was run, there is way to "focus" them. This is sort of related to the point above, but there are structural ways it could have been prevented even without editing/collaboration software. For instance, they could have had a better categorization system in order to head off some of the overlap.

Third, there were tons of complaints about the selection process for the 1st round finalists - it was done by a "diverse group of experts", i.e. without involvement of anyone outside the organization. Since that round took the number of ideas from 22,000+ to 70, it was obviously a brutal job - and I bet that given a slightly more democratic system where e.g. people can "recommend" an idea, the results would even have been better. The complaints also point out that the sponsoring organization (a union) biased the round one winners to ideas they approved of (unsurprisingly). I'd say that basically they failed to use the power of the internet to help with the project beyond the submission stage.

And finally, the contest is impotent. Despite the awards, it's clear that nothing will come from the entire thing, and that's rather dispiriting. Think how much more interesting it would be if the contest was held by the ruling political party or the president? Or even by an electoral campaign? What about if the awarding organization was going to spend $10M lobbying for the ideas in Washington?

If only those in Government had a little more vision, this kind of thing could be hugely transformational. For instance, how about this idea: the winner of the quarterly contest becomes a new "elected" member of "the idea congress", responsible for implementing their idea. 4-year term, it's a sitting body of 16 people (4 quarters * 4 years = 16) with a billion dollar budget to implement the winning ideas. Each sitting member has a staff, etc. It does not replace congress or the president, but rather supplements them with a constant stream of citizen approved ideas. The billion dollars can of course be supplemented with more if they can convince congress of the important and value of the idea...