Tuesday, January 29, 2013

Tuesday Links

  • Above is Chinese coal consumption vs the rest of the world.  Read it and weep.
  • New York Time is cautiously opening up to the idea that being prepared for a risk of major disaster may not be crazy.  I'm in agreement.  While I think the risk of a serious civilizational collapse is small, I don't think it's zero.  And certainly the risk of living through major regional weather disasters is just going to keep getting higher.
  • A philosopher buys into the singularity as an existential risk.
  • Maybe we should rethink our concerns about nuclear waste.  Apparently it's not nearly as bad for wildlife as humans are.
  • Cellulosic ethanol fail (although I'd really love to see Robert Rapier on a unicorn - somebody should totally get on that with Photoshop).


Stephen B. said...

I think that the likes of The New York Times and other groups that were initially hostile to the the idea of "Preppers" have nobody to blame but themselves for creating images of strange, mainly right wing political people preparing for some extreme event. If only such doubters were a little bit older, older enough to have parents or grandparents that lived through the Depression, or who had to live through bad weather such as blizzards or hurricanes before modern response teams, infrastructure, and support networks came into being, then perhaps the former would understand what being prepared is really all about.

I look at being prepared as merely doing what is prudent and smart - doing what people often did for themselves and their families before the modern, just-in-time inventory way of doing business inevitably worked its way into our personal, family lives - to the latter's detriment.

Some of us always understood that we all should be as prepared for bad times as is reasonably possible, if only to help out the entire community by not being an undue burden to the rest of the recovery efforts, whatever the nature of the disaster. It's not about being a political kook or having so much doubt in our government as to have open contempt for it. Rather, it's being responsible, pro-community, and smart.

I guess the likes of The Times and other, "progressive" people are finally figuring this out. It's about time.

Stephen B. said...

As for wildlife returning to the exclusion zone around Chernobyl, the whole thing reminds me of the woman who supposedly rode through the area on her motorcycle, taking pictures to document the area. As most here probably know, the woman behind the www.kidofspeed.com website perpetrated one of the Internet's better known hoaxes, as there was no actual motorcycle trip. She, or somebody, had merely lifted photographs from various Soviet and Russian sources. Nevertheless, that website and the very real photographs contained within got me thinking several years ago about the whole impact of a large scale radiation release on the natural environment and indeed, wildlife on the whole, seem to thrive.

Still, there are many questions.

While species such as wolves and other large mammals have done well, are we sure individual animals haven't suffered, meaning either painful lives, and/or living very short lives? I'm guessing that reproduction within a species is keeping ahead of increased individual mortality and mutation. That might be an acceptable outcome for a population of wolves, but probably not for a city full of humans. For that matter, all I've really read about is the proliferation of wolves.

As for the plant life, I suppose there are now ongoing studies, but I haven't seen anything in the popular press. Anybody know more?

The evacuation does show, however, that many aspects of the natural world seem to quickly flow back into any void humanity has vacated, but closer study is probably indicated.

Stephen B. said...

My apologies. I got the URL of the supposed motorcycle trip with the photographs wrong. It's http://www.kiddofspeed.com/ (with two d's) and it's still up and running.

Greg said...

I have trouble seeing the singularity as a looming risk to civilisation.

I think that Ray Kurzweil and his followers confuse invention and innovation: the initial demonstration of a thing and its adoption in practice.

Innovation takes place a lot more slowly than invention. For example, Michael Faraday invented the electric motor in the 1820s, but the first adoption of for practical use took place in the 1870s, and it wasn't until the 1900s that they really started to displace steam power in factories.

One AI will have as much power to change things as one electric motor, one H-bomb...or, possibly, one gallon of cellulosic ethanol.

In the near term (the next fifty years), I'm more worried by two other things: topsoil loss through desertification, and the decline in GDP per capita in the advanced countries.

Topsoil loss is covered at length (but perhaps unsystematically) by Lester Brown and others. Loss of some topsoil puts pressure on the remaining soil: a classic non-linear situation.

The trend* in per-capita GDP growth in advanced countries has been steadily down over the last fifty years. If the trend continues, OECD GDP per capita will start to shrink around 2020 or 2030.

That seems like a situation that could turn non-linear, too.

* FRED is the data repository maintained by the Federal Reserve Bank of St. Louis; the series used are its Real GDP (USD million) and Population (000) series for each of the countries. Fifty years is long enough for a trend, and the simplest linear model appears to fit the data well. Economists (still) aren't agreed on any more complicated model.

Stuart Staniford said...

Greg - your argument about 1 AI doesn't make sense - it's a set of algorithm, when you have one, you can make millions in no time. And adoption of new software technologies these days takes of order one decade (from university research papers to widespread practical use).

As to soil loss, I looked into it briefly a number of years back and was unable to convince myself there was an urgent global problem. I think Lester Brown is biassed way alarmist.


Greg said...


yes, once you have one AI running on a von Neumann architecture you could in principle make millions or billions of them quite quickly.

But you need to have a use for them, some-one to buy them, or they won't get made.

It's not clear to me that full AIs would offer much value over more specialised tools: brute-force machine learning systems (descendants in spirit of IBM's Watson) and limited autonomous systems such as self-driving cars and descendants of the "teach-by-guiding" robot we saw a few weeks ago.

It will take time for investors and regulators to be convinced about the benefits and risks of AIs once one has been demonstrated. (One risk: do they develop mental illnesses? How frequently and predictably? Are there warning signs? Are they treatable? Another: how will employees and customers react to interacting with AIs? How will the AIs react to them? How will customers react to the knowledge that our products are made by AIs and not humans -- will AI be the new GM or nuclear power?) These are not word-processing programs or CNC machine tools.

There are more fundamental risks. How will the structures of businesses and markets change once AIs are available? I don't think that anyone has much idea at present, and I think that finding good answers is going to take time to experiment. I expect "the singularity" to take several decades to play out, not several months or years. And it might stall.

Greg said...

On topsoil, I wondered whether you had revised your opinion since early '08, in the light of Aiguo Dai's projections (your Future of Drought series) and continued acceleration in carbon emissions.

It's not now a great stretch to imagine re-formation of the North American Desert--transiently, for a few hundred years, until the ocean has warmed up and Pliocene-type conditions reassert themselves. Nor to imagine that, putting a North American desert together with persistent drought in southern Europe, South America, South-East Asia and Australia, we get a decline in food production in these areas.

That would put pressure on the remaining food-growing areas to grow more, and bring pasture into crop production. (Especially if US and EU biofuel policies aren't reversed.) These areas themselves are likely to be stressed by rainfall variability.

In theory there is plenty of room for yield improvement in Russia/Ukraine, South and East Asia and Africa, despite increased water stress.

In practice, the current problems in those places seem likely to persist: lack of secure property rights and fair credit markets for farmers, infrastructure and support industry weaknesses, and poor crop markets (monopoly buyers). Those problems reduce the incentive or ability to grow more than current amounts.

If deserts do expand suddenly, we could be in for an interesting time.

sunbeam said...

To look at the "One AI" thing from a different angle, exactly how many AI learning systems do you need exactly?

I'm not qualified to answer that, but it seems to me that even one could conceivably shake up a lot of things.

And on a slightly related note, I have to wonder if robots might become very cheap. Consider that with advances in wireless communication, and other means of information transport, I can easily imagine banks of servers in New York or somewhere controlling agricultural robots in Iowa, Nebraska, India, or where have you.

Just saying I can easily imagine some sort of rudimentary processor, a comm unit which would basically be a cell phone, and whatever electromechanical apparatus and sensor arrays that let it do work. Heck the comm unit could also be the processor. Just stick one of whatever the latest cell phone is in a hookup harness and you are ready to go.

I can't speculate on the hardware requirements but it is again easy to imagine one server controlling a number of different robots simultaneously.

Unknown said...

Regarding the return of wildlife in the Chernobyl exclusion zone, one should be careful to extrapolate on the actual effects simply by looking at a few pictures of animals.

I found at least one other study that shows a significant decrease in biodiversity in the exclusion zone.

I also recall another study, which I can't find, that compared the biodiversity, fauna and flora health in the exclusion zone to a nature preserve in Russia, with similar climate and ecosystem, that was not affected by Chernobyl. It found that plants and animals in the exclusion zone bore a significantly higher level of birth defects and genetic abnormalities than their unaffected counterparts. Biodiversity was also markedly lower.

The conclusion is not that radioactive fallout is not as bad as we thought, it's that human presence is much worse.

Manolo said...

Hi Stuart !

Regarding coal consumption, wildfires and other nastiness, here is a good cause that merits ample support and will help us understand more about the unbelievable rate of change we witness in the Arctic:


Just to add: the tremendous ice-loss on land makes the entire system up there very unstable. Land rises. Might even explain while the magnetic north pole has been on a race in recent years and the many earth quakes in the Arctic and volcano eruptions of long thought extinct volcanoes.

Something is definitively brewing up there !