Yeah. It’s been one of my mantras since before the dotcom boom that there is a world of difference between the marginal cost of something falling to near zero and it falling to actually zero.
I was going to use exactly that sentence! but of course, ignoring things that seem to be nearly zero is pretty much essential for analysis of any kind, or thinking at all. I said in a talk a few months ago that "the only people who really try to pay attention to absolutely every piece of information they have are babies"
Weightlessness was always silly, the sort of thing you have to be highly educated to believe--which was why the Economist was shilling it relentlessly back in the day (is it thirty years ago?). My reaction to "dematerialisation" was, I'll believe it when I see wealthy people living in twenty kilogramme houses.
We _were_ in danger of entering a post-Moore's Law era when even Microsoft couldn't sufficiently bloat their software to slow down a five-year-old machine and we could perhaps begin to un-fuck computer science as a discipline and work out what we could do with less power, streamline some of our tottering pile of abstractions, create interfaces that didn't suck etc. etc.
Luckily for the Shovel Salesmen they discovered a new, special, type of gold that could only be mined with the latest and greatest shovels just at the exact convenient moment. In retrospect it was staggeringly naïve of me to expect anything else.
“Practically everything we do, from eating an ice to crossing the Atlantic, and from baking a loaf to writing a novel, involves the use of coal, directly or indirectly. For all the arts of peace coal is needed; if war breaks out it is needed all the more. In time of revolution the miner must go on working or the revolution must stop, for revolution as much as reaction needs coal. Whatever may be happening on the surface, the hacking and shovelling have got to continue without a pause, or at any rate without pausing for more than a few weeks at the most. In order that Hitler may march the goose-step, that the Pope may denounce Bolshevism, that the cricket crowds may assemble at Lords, that the poets may scratch one another's backs, coal has got to be forthcoming. But on the whole we are not aware of it; we all know that we ‘must have coal’, but we seldom or never remember what coal-getting involves. Here am I sitting writing in front of my comfortable coal fire. It is April but I still need a fire. Once a fortnight the coal cart drives up to the door and men in leather jerkins carry the coal indoors in stout sacks smelling of tar and shoot it clanking into the coal-hole under the stairs. It is only very rarely, when I make a definite mental-effort, that I connect this coal with that far-off labour in the mines. It is just ‘coal’ — something that I have got to have; black stuff that arrives mysteriously from nowhere in particular, like manna except that you have to pay for it.”
- George Orwell, The Road to Wigan Pier
I share this both because it is the same pattern or systems that we see time and again, and now in AI’s material constraints, and *also* because reading this in 2024, you can’t help but think, “well, I suppose if we can break out of today’s energy-suffering accounting with some bets on breakthroughs, more use today doesn’t seem horribly wrong.”
«“It is just ‘coal’ — something that I have got to have; black stuff that arrives mysteriously from nowhere in particular, like manna except that you have to pay for it.”»
Jevons wrote about that and more recently Wrigley:
I once had a conversation with a Serious Economist who stated authoritatively that coal/oil are not important to The Economy because they account for only a few percent points of GDP in developed countries and "intellectual property, professional services, brands" accounted for far more.
To me he seemed to disregard the notion of "consumer surplus" and taking that into account I think that they actually deliver 90% of GDP in developed countries.
The fundamental brittleness of logic, and from that machine logic creates a form of 'mass' (as in inertia). The result is that we're more on the road to 'complexity crunch' than a 'singularity point'. A second fundamental issue is that working in the discrete (digital) realm has a lot of strengths, but a fundamental weakness: the world is not Q, it is R (the world is not rational, it is real). Trying to do 'real' things with 'rational' systems inevitably leads to an infinite scaling problem, which overcomes any Moore's Law-like productivity growth we see. Besides, it seems many have mistaken the start of an S-curve for the start of an exponential one.
«Electricity production in the US and other developed countries has barely moved in the past 20 years.»
One of the small details that is rarely mentioned because of its unimportance :-) is that electricity *consumption* per person has been collapsing rather quickly in "the west" since a few years *before* the 2008 GFC, breaking the long term correlation between that and reported "GDP"; that had never happened before except during the collapse of eastern European economies after their defeat in 1991 (plus 1980-1985 in the UK as Thatcher worked her magic on the UK economy).
«urgent planning problems of the location of datacenters and occasional shortages of the water needed to cool them. [...] build just a leetle nuclear power plant, just for them. Well maybe not such a little plant. And maybe not just the one …»
While factories, research labs, offices, have been relentlessly moved offshore to jurisdictions where labor unions and strikes are illegal or powerless, data centres are being built in "the west" for the following reasons:
* They employ very few people.
* They get protected by the military and police forces of "the west" where the governments have been thoroughly "sponsored" by the owners of those data centers.
* By putting the data centers in (usually remote areas of) one ("developed") country and the people who work on those computers (programmers, researchers, data entry clerks, call centre operators, ...) in another ("developing") country it is possible to avoid any risk of strikes and occupations of those data centers by those workers, and it is possible to "lock-out" potential strikers with a few clicks of the mouse (by disabling their accounts) and replace them quickly with scabs from another ("developing") country. with another few mouse clicks.
«what happened to the weightless economy? I mean that figuratively rather than literally; all the things that Danny Quah was talking about in the article of that title (intellectual property, professional services, brands)»
That "weightless economy" was merely an euphemism for what I call "headquarterisation" which was a crafty bit of propaganda that went like this:
* Do not worry about offshoring because...
* Only the polluting factories will be sent offshore with low paid jobs to be done by gullible idiots in hellhole countries.
* We will keep all the headquarters with paper-pushing "weightless" jobs and our country will be a beautiful park with mansions for everybody as everybody will just process "data" and "information" as everybody will be a highly paid Queen's Counsel, Marketing Planner, Head of Technology, Project Manager, Top Designer, Chief Financial Officer, Consultant for Strategy, Vice-President, Lead Scientist, because the number of jobs in these upper-middle class roles will be directly proportional to the number of graduates we generate (as that guy said "education, education, education").
* None of us will need to keep our mansions clean and our green and pleasant land gardened because all the non-headquarters service jobs will be done by gullible idiots who will immigrate to our country from the same hellhole countries.
The overall assumption of that propaganda was that those gullible idiots in hellhole countries would want to work real hard for very low wages to make highly useful products for cheap that then we would sell back to them at high prices (thanks to our sole control of "intellectual property, professional services, brands") with the large difference between the two being pocketed by our headquarters, as in Japan for example, because those gullible idiots in hellhole countries would not have the cleverness or ability to setup their own businesses with their own headquarters. Indeed even today RCA televisions, RACAL cellphones, Leyland cars, Digital Equipment computers dominate the world markets. :-)
Yeah. It’s been one of my mantras since before the dotcom boom that there is a world of difference between the marginal cost of something falling to near zero and it falling to actually zero.
I was going to use exactly that sentence! but of course, ignoring things that seem to be nearly zero is pretty much essential for analysis of any kind, or thinking at all. I said in a talk a few months ago that "the only people who really try to pay attention to absolutely every piece of information they have are babies"
Weightlessness was always silly, the sort of thing you have to be highly educated to believe--which was why the Economist was shilling it relentlessly back in the day (is it thirty years ago?). My reaction to "dematerialisation" was, I'll believe it when I see wealthy people living in twenty kilogramme houses.
We _were_ in danger of entering a post-Moore's Law era when even Microsoft couldn't sufficiently bloat their software to slow down a five-year-old machine and we could perhaps begin to un-fuck computer science as a discipline and work out what we could do with less power, streamline some of our tottering pile of abstractions, create interfaces that didn't suck etc. etc.
Luckily for the Shovel Salesmen they discovered a new, special, type of gold that could only be mined with the latest and greatest shovels just at the exact convenient moment. In retrospect it was staggeringly naïve of me to expect anything else.
“Practically everything we do, from eating an ice to crossing the Atlantic, and from baking a loaf to writing a novel, involves the use of coal, directly or indirectly. For all the arts of peace coal is needed; if war breaks out it is needed all the more. In time of revolution the miner must go on working or the revolution must stop, for revolution as much as reaction needs coal. Whatever may be happening on the surface, the hacking and shovelling have got to continue without a pause, or at any rate without pausing for more than a few weeks at the most. In order that Hitler may march the goose-step, that the Pope may denounce Bolshevism, that the cricket crowds may assemble at Lords, that the poets may scratch one another's backs, coal has got to be forthcoming. But on the whole we are not aware of it; we all know that we ‘must have coal’, but we seldom or never remember what coal-getting involves. Here am I sitting writing in front of my comfortable coal fire. It is April but I still need a fire. Once a fortnight the coal cart drives up to the door and men in leather jerkins carry the coal indoors in stout sacks smelling of tar and shoot it clanking into the coal-hole under the stairs. It is only very rarely, when I make a definite mental-effort, that I connect this coal with that far-off labour in the mines. It is just ‘coal’ — something that I have got to have; black stuff that arrives mysteriously from nowhere in particular, like manna except that you have to pay for it.”
- George Orwell, The Road to Wigan Pier
I share this both because it is the same pattern or systems that we see time and again, and now in AI’s material constraints, and *also* because reading this in 2024, you can’t help but think, “well, I suppose if we can break out of today’s energy-suffering accounting with some bets on breakthroughs, more use today doesn’t seem horribly wrong.”
«“It is just ‘coal’ — something that I have got to have; black stuff that arrives mysteriously from nowhere in particular, like manna except that you have to pay for it.”»
Jevons wrote about that and more recently Wrigley:
https://royalsocietypublishing.org/doi/10.1098/rsta.2011.0568
https://oll.libertyfund.org/titles/jevons-the-coal-question
I once had a conversation with a Serious Economist who stated authoritatively that coal/oil are not important to The Economy because they account for only a few percent points of GDP in developed countries and "intellectual property, professional services, brands" accounted for far more.
To me he seemed to disregard the notion of "consumer surplus" and taking that into account I think that they actually deliver 90% of GDP in developed countries.
The fundamental brittleness of logic, and from that machine logic creates a form of 'mass' (as in inertia). The result is that we're more on the road to 'complexity crunch' than a 'singularity point'. A second fundamental issue is that working in the discrete (digital) realm has a lot of strengths, but a fundamental weakness: the world is not Q, it is R (the world is not rational, it is real). Trying to do 'real' things with 'rational' systems inevitably leads to an infinite scaling problem, which overcomes any Moore's Law-like productivity growth we see. Besides, it seems many have mistaken the start of an S-curve for the start of an exponential one.
https://ea.rna.nl/2024/09/28/like-we-dont-see-air-we-dont-see-the-digital-revolution/
https://ea.rna.nl/2024/11/10/hello-human-intelligence-meet-complexity-crunch/
The power demands of data centres have been greatly overstated by nuclear fans, just ast PC demand was overstated by coal fans decades ago.
https://www.forbes.com/forbes/1999/0531/6311070a.html
Electricity production in the US and other developed countries has barely moved in the past 20 years.
https://www.iea.org/countries/united-states/electricity
https://reneweconomy.com.au/ai-wont-use-as-much-electricity-as-we-are-told-and-its-not-a-reason-to-slow-transition-to-renewables/
«Electricity production in the US and other developed countries has barely moved in the past 20 years.»
One of the small details that is rarely mentioned because of its unimportance :-) is that electricity *consumption* per person has been collapsing rather quickly in "the west" since a few years *before* the 2008 GFC, breaking the long term correlation between that and reported "GDP"; that had never happened before except during the collapse of eastern European economies after their defeat in 1991 (plus 1980-1985 in the UK as Thatcher worked her magic on the UK economy).
https://blissex.wordpress.com/wp-content/uploads/2018/02/dataelectreuothersconsperhead1960to2015.png
https://blissex.wordpress.com/wp-content/uploads/2018/02/dataenergyeuuschjppergdpperhead1960to2014.png
«urgent planning problems of the location of datacenters and occasional shortages of the water needed to cool them. [...] build just a leetle nuclear power plant, just for them. Well maybe not such a little plant. And maybe not just the one …»
While factories, research labs, offices, have been relentlessly moved offshore to jurisdictions where labor unions and strikes are illegal or powerless, data centres are being built in "the west" for the following reasons:
* They employ very few people.
* They get protected by the military and police forces of "the west" where the governments have been thoroughly "sponsored" by the owners of those data centers.
* By putting the data centers in (usually remote areas of) one ("developed") country and the people who work on those computers (programmers, researchers, data entry clerks, call centre operators, ...) in another ("developing") country it is possible to avoid any risk of strikes and occupations of those data centers by those workers, and it is possible to "lock-out" potential strikers with a few clicks of the mouse (by disabling their accounts) and replace them quickly with scabs from another ("developing") country. with another few mouse clicks.
«what happened to the weightless economy? I mean that figuratively rather than literally; all the things that Danny Quah was talking about in the article of that title (intellectual property, professional services, brands)»
That "weightless economy" was merely an euphemism for what I call "headquarterisation" which was a crafty bit of propaganda that went like this:
* Do not worry about offshoring because...
* Only the polluting factories will be sent offshore with low paid jobs to be done by gullible idiots in hellhole countries.
* We will keep all the headquarters with paper-pushing "weightless" jobs and our country will be a beautiful park with mansions for everybody as everybody will just process "data" and "information" as everybody will be a highly paid Queen's Counsel, Marketing Planner, Head of Technology, Project Manager, Top Designer, Chief Financial Officer, Consultant for Strategy, Vice-President, Lead Scientist, because the number of jobs in these upper-middle class roles will be directly proportional to the number of graduates we generate (as that guy said "education, education, education").
* None of us will need to keep our mansions clean and our green and pleasant land gardened because all the non-headquarters service jobs will be done by gullible idiots who will immigrate to our country from the same hellhole countries.
The overall assumption of that propaganda was that those gullible idiots in hellhole countries would want to work real hard for very low wages to make highly useful products for cheap that then we would sell back to them at high prices (thanks to our sole control of "intellectual property, professional services, brands") with the large difference between the two being pocketed by our headquarters, as in Japan for example, because those gullible idiots in hellhole countries would not have the cleverness or ability to setup their own businesses with their own headquarters. Indeed even today RCA televisions, RACAL cellphones, Leyland cars, Digital Equipment computers dominate the world markets. :-)
The mill reference reminds me of Brian Merchant's Blood in the Machine