The land around New York City is worth a lot. A 2008 analysis estimated prices for land, not counting buildings etc., for most (~80%?) of the nearby area (2750 square miles, = a 52 mile square). The total New York area land value (total land times ave price) was 5.5T$ (trillion) in 2002 and 28T$ in 2006.
"limited fooms due to architectual insights" sounds like saying there have been a few unusually lumpy innovations, but most innovation have come from many small gains. Which is what I say.
A lot of Jane Jacobs's writing is very specifically about a city-ularity, how this is bad and we should be putting a lot of thought into things like optimal currency zones and distinguishing growth via real production from extractive growth and wireheading.
I'll bite the bullet. Before computers, the city was the main paradigm for automation technology among humans. There have been limited FOOMs due to architectural insights - the British Empire, and the world of multinational corporations that emerged from it, is a recent example: https://www.ribbonfarm.com/...
This has been a mixed result for humanity and created dangerous path dependencies much less under the control of free humans than prior arrangements were. The world really is running much more like a single city than it used to in ways that we should be very worried about (see here for a simple example you came up with: https://www.overcomingbias.... ).
Location values are about 1/3 to 1/2 of total real estate value nationwide, but that 'land share' ratio increases to above 4/5 of total real estate value in New York City and even 9/10 of total value in San Francisco.
Karl, the productivity potential of a better AI design also can not be realized without it being embodied in hardware and without many others being induced to interact with it to solve their many problems. By itself a mere design for an AI is no more useful than a mere design for a better city.
1) The city requires new physical capital to work. Its not enough to have a perfect city design. Millions of tons of matter have to be rearranged in order for that design to work. An AI by contrast only has to rearrange a small amount of matter to work.
This implies that much bolder steps can be taken. Creating a new world class city from scratch, as Dubai is attempting, is no small feat. And getting it wrong means the loss of enormous wealth. Thus even if we have some pretty good ideas on city design they will only come into being incrementally. A huge city-leap is unlikely.
2) The value from a city comes from the people living in it. The transactions cost associated with moving the entire population of a city are enormous. So, even if I built New York II that was way better than New York it is likely that it will only fill up slowly and I will only get super rich slowly. Thus the run away potential is lower.
3) While a great city increases ones potential to create greater cities, there is still the limit of human cognition. Yes, we do better thinking in cities but I am not sure if it is so much better than the thinking between cities. This implies that the intracity advantage to creating even better cities is not that great and so the first mover advantage is lessoned.
In short cities do not have the property that make the AI singularity so dangerous: cheap, strongly self-reinforcing replication.
Land value is also affected by whether regulation restricts higher value uses, thus generating extra scarcity. The difference between the Zoned Zone and Flatland, as Krugman put it.
It seems much more obvious to me that city planning consists of getting lots of details right than it does that AI does.
But I think the biggest difference between cities and AI is that you don't have to attract millions of people to your AI to make it work. Even if the analogy would work without this difference, it's quite hard to separate this difference from your intuitions, making the analogy not too useful.
"It’s a little off-topic, but can anyone explain what’s so special about the location aside from the fact that’s it’s where New York city actually is?"
It's a significantly shorter journey from Europe to Boston or NYC than it is to say Philly or Baltimore. And Southern ports didn't have the human or economic capital (nor the climate) to grow like the north did.
"The Hudson river is big and carries a lot of traffic, but nowhere near as much as the Mississippi river. "
What about 200 years ago? 100 years ago?The Lower Mississippi watershed is a dump, mostly inhabited by people of low intelligence, and it lacks similar quantities of farmland. The Upper Mississippi and its tributaries were rendered relatively insignificant by the time that their (vast and rich) agricultural lands were put into serious use, because of rail. Shipping goods down to New Orleans would be like shipping them from New York to Greenland... what would be the point? Goods (and people) went by rail east (to New York) and west instead.
Besides, traffic through the Hudson can go *to* the Mississippi, thanks to canals. The Erie canal sealed New York's place as the dominant east coast city, as NYC became a port to the Great Lakes watershed. (Undercutting Montreal as well for that matter) Chicago's canal to the Mississippi continued the waterway from NYC to the Mississippi. By 1900 or so, New York and Chicago were the two great transportation centers of the nation.
"NY is a little closer to Europe, but not that much, really – maybe it was a factor in the sailing days, but those are long gone."
Regardless of whether or not the factor disappears, the effects of it remain, with human and economic capital already put in place.
"I suspect there’s a lot of lock-in from the days before air conditioning,"
Indeed... the only reason the South has been able to grow in recent decades is air conditioning. There are other factors making it grow, but they'd all be inhibited by lack of AC.
Engineered-from-scratch machine intelligence and intelligence augmentation of humans are partner technologies. At least they have been so far.
You simply use machines to pre-process human sensory inputs and post-process their motor outputs (e.g. keyboards, mice, monitors, speakers) So: better machine intelligence leads quickly to better machine-augmented humans.
I think that the proximity to Niagara Falls might have helped New York substantially too back in the days when the vast majority of electricity was created by dirty coal burning plants very near to or inside of cities. This probably enabled a greater concentration of industry at a given level smog than other cities could attain. Toronto is Canada's largest city after all, I wonder how much that helped.
Yes, New Songdo demonstrates that it is not that capital can't be collected to fund a new city from scratch, nor that its builder's can't collect much of the gains from a wise investment. It is that we just don't know how to make dramatically better cities.
I agree that there are two distinct issues: the suddenness of the jump one team may have over any other, and the relative advantage of clean-slate explicitly-coded AI over incremental modifications of human intelligence. I have related reasons to be skeptical about each of them.
Also cannot we see Dubai as an Evil Unfriendly city-ularity with a winning formula?
Geographical "advantage": Ahemmm, not a drop of water, a hot desert with periodical dust storms. BUT sits on or rather near the biggest oil-fields on the planet, with investors and sovereign wealth funds sitting on humongous piles of petrodollars that have to chase every hare-brained investment because all the good ones have been snapped up already.
An immoral "government" (Royals treating their country as a private fiefdom out of greed and only motivated by money.)
Built by near slaves from India tricked by promises of riches.Run by foreign ex-pats with no roots in the city and only there for the money.
Now this model kind of hit a speed bump and shown some vulnerability with the financial crises. Yet it will recover and is already frankly.
An apparent successful unfriendly city-ularity with no apparent morals or ethics (except that of profit maximizing).
Yet am I afraid of Dubai taking over the world?HAHA
Hmm why would you need super smarties for your city theory... i think the success of the most successful cities is not that hard, there is three (or four) main points.
1: Time of foundation.2: Some unique geographic location.3: Success of the political-economic union in which it resides.4: There can also be some unique quirk or historical reason.
Regarding New York, it was founded early, was / is the main eastern port for the flow of massive immigration that made America (remember most immigrants saw the Statue of Liberty as they arrived at their new homeland). It was and is in effect the capital of the USA (A very successful economic union). Combine that with the island location which results in the way out of whack land prices.(Simple supply and demand, NYC's unique quirk).
Another example is Cape Town and the Gauteng province (In effect a fusion of 3 big cities, also known as the PWV triangle and the richest) in South Africa.
Cape Town was the first town and port, and now one of the biggest cities. The Gauteng province sits in the middle of the country no ports, dry location (but has significant rivers). Almost all other mega cities are at the coast. The reason is that this weird location was the richest gold field when discovered 150 years ago during the Gold Rush and the export of the Industrial Revolution by England to its colonies.(And this can over ride the first movers, Gauteng is bigger and richer than Cape Town even when it is 200 years younger.)
Knowing this it is still impossible to spark a city-ularity, because the process is so evolutionary and organic.
Regarding AI it is going to be a lot more evolutionary than revolutionary. It is not going to take over the world in one fell swoop.
The city-ularity is a good example at debunking the singularity.
Re: "There’s a scenario you don’t seem to be considering, which is a gradual, world-wide development of deliberately designed AIs, leading eventually (over the very long term) to a “singularity”, a lifestyle beyond which it is difficult to make predictions."
Agreed, except: if a “singularity”, is "a lifestyle beyond which it is difficult to make predictions" - haven't we had some of those already? 2000 was surely 1800's "singularity" on those grounds. Since such terminology is so vague and silly, why perpetuate its use?
"limited fooms due to architectual insights" sounds like saying there have been a few unusually lumpy innovations, but most innovation have come from many small gains. Which is what I say.
A lot of Jane Jacobs's writing is very specifically about a city-ularity, how this is bad and we should be putting a lot of thought into things like optimal currency zones and distinguishing growth via real production from extractive growth and wireheading.
I'll bite the bullet. Before computers, the city was the main paradigm for automation technology among humans. There have been limited FOOMs due to architectural insights - the British Empire, and the world of multinational corporations that emerged from it, is a recent example: https://www.ribbonfarm.com/...
This has been a mixed result for humanity and created dangerous path dependencies much less under the control of free humans than prior arrangements were. The world really is running much more like a single city than it used to in ways that we should be very worried about (see here for a simple example you came up with: https://www.overcomingbias.... ).
Location values are about 1/3 to 1/2 of total real estate value nationwide, but that 'land share' ratio increases to above 4/5 of total real estate value in New York City and even 9/10 of total value in San Francisco.
Karl, the productivity potential of a better AI design also can not be realized without it being embodied in hardware and without many others being induced to interact with it to solve their many problems. By itself a mere design for an AI is no more useful than a mere design for a better city.
There are two key difference with the city
1) The city requires new physical capital to work. Its not enough to have a perfect city design. Millions of tons of matter have to be rearranged in order for that design to work. An AI by contrast only has to rearrange a small amount of matter to work.
This implies that much bolder steps can be taken. Creating a new world class city from scratch, as Dubai is attempting, is no small feat. And getting it wrong means the loss of enormous wealth. Thus even if we have some pretty good ideas on city design they will only come into being incrementally. A huge city-leap is unlikely.
2) The value from a city comes from the people living in it. The transactions cost associated with moving the entire population of a city are enormous. So, even if I built New York II that was way better than New York it is likely that it will only fill up slowly and I will only get super rich slowly. Thus the run away potential is lower.
3) While a great city increases ones potential to create greater cities, there is still the limit of human cognition. Yes, we do better thinking in cities but I am not sure if it is so much better than the thinking between cities. This implies that the intracity advantage to creating even better cities is not that great and so the first mover advantage is lessoned.
In short cities do not have the property that make the AI singularity so dangerous: cheap, strongly self-reinforcing replication.
Land value is also affected by whether regulation restricts higher value uses, thus generating extra scarcity. The difference between the Zoned Zone and Flatland, as Krugman put it.
It seems much more obvious to me that city planning consists of getting lots of details right than it does that AI does.
But I think the biggest difference between cities and AI is that you don't have to attract millions of people to your AI to make it work. Even if the analogy would work without this difference, it's quite hard to separate this difference from your intuitions, making the analogy not too useful.
"It’s a little off-topic, but can anyone explain what’s so special about the location aside from the fact that’s it’s where New York city actually is?"
It's a significantly shorter journey from Europe to Boston or NYC than it is to say Philly or Baltimore. And Southern ports didn't have the human or economic capital (nor the climate) to grow like the north did.
"The Hudson river is big and carries a lot of traffic, but nowhere near as much as the Mississippi river. "
What about 200 years ago? 100 years ago?The Lower Mississippi watershed is a dump, mostly inhabited by people of low intelligence, and it lacks similar quantities of farmland. The Upper Mississippi and its tributaries were rendered relatively insignificant by the time that their (vast and rich) agricultural lands were put into serious use, because of rail. Shipping goods down to New Orleans would be like shipping them from New York to Greenland... what would be the point? Goods (and people) went by rail east (to New York) and west instead.
Besides, traffic through the Hudson can go *to* the Mississippi, thanks to canals. The Erie canal sealed New York's place as the dominant east coast city, as NYC became a port to the Great Lakes watershed. (Undercutting Montreal as well for that matter) Chicago's canal to the Mississippi continued the waterway from NYC to the Mississippi. By 1900 or so, New York and Chicago were the two great transportation centers of the nation.
"NY is a little closer to Europe, but not that much, really – maybe it was a factor in the sailing days, but those are long gone."
Regardless of whether or not the factor disappears, the effects of it remain, with human and economic capital already put in place.
"I suspect there’s a lot of lock-in from the days before air conditioning,"
Indeed... the only reason the South has been able to grow in recent decades is air conditioning. There are other factors making it grow, but they'd all be inhibited by lack of AC.
Engineered-from-scratch machine intelligence and intelligence augmentation of humans are partner technologies. At least they have been so far.
You simply use machines to pre-process human sensory inputs and post-process their motor outputs (e.g. keyboards, mice, monitors, speakers) So: better machine intelligence leads quickly to better machine-augmented humans.
I think that the proximity to Niagara Falls might have helped New York substantially too back in the days when the vast majority of electricity was created by dirty coal burning plants very near to or inside of cities. This probably enabled a greater concentration of industry at a given level smog than other cities could attain. Toronto is Canada's largest city after all, I wonder how much that helped.
Yes, New Songdo demonstrates that it is not that capital can't be collected to fund a new city from scratch, nor that its builder's can't collect much of the gains from a wise investment. It is that we just don't know how to make dramatically better cities.
I agree that there are two distinct issues: the suddenness of the jump one team may have over any other, and the relative advantage of clean-slate explicitly-coded AI over incremental modifications of human intelligence. I have related reasons to be skeptical about each of them.
Also cannot we see Dubai as an Evil Unfriendly city-ularity with a winning formula?
Geographical "advantage": Ahemmm, not a drop of water, a hot desert with periodical dust storms. BUT sits on or rather near the biggest oil-fields on the planet, with investors and sovereign wealth funds sitting on humongous piles of petrodollars that have to chase every hare-brained investment because all the good ones have been snapped up already.
An immoral "government" (Royals treating their country as a private fiefdom out of greed and only motivated by money.)
Built by near slaves from India tricked by promises of riches.Run by foreign ex-pats with no roots in the city and only there for the money.
Now this model kind of hit a speed bump and shown some vulnerability with the financial crises. Yet it will recover and is already frankly.
An apparent successful unfriendly city-ularity with no apparent morals or ethics (except that of profit maximizing).
Yet am I afraid of Dubai taking over the world?HAHA
Hmm why would you need super smarties for your city theory... i think the success of the most successful cities is not that hard, there is three (or four) main points.
1: Time of foundation.2: Some unique geographic location.3: Success of the political-economic union in which it resides.4: There can also be some unique quirk or historical reason.
Regarding New York, it was founded early, was / is the main eastern port for the flow of massive immigration that made America (remember most immigrants saw the Statue of Liberty as they arrived at their new homeland). It was and is in effect the capital of the USA (A very successful economic union). Combine that with the island location which results in the way out of whack land prices.(Simple supply and demand, NYC's unique quirk).
Another example is Cape Town and the Gauteng province (In effect a fusion of 3 big cities, also known as the PWV triangle and the richest) in South Africa.
Cape Town was the first town and port, and now one of the biggest cities. The Gauteng province sits in the middle of the country no ports, dry location (but has significant rivers). Almost all other mega cities are at the coast. The reason is that this weird location was the richest gold field when discovered 150 years ago during the Gold Rush and the export of the Industrial Revolution by England to its colonies.(And this can over ride the first movers, Gauteng is bigger and richer than Cape Town even when it is 200 years younger.)
Knowing this it is still impossible to spark a city-ularity, because the process is so evolutionary and organic.
Regarding AI it is going to be a lot more evolutionary than revolutionary. It is not going to take over the world in one fell swoop.
The city-ularity is a good example at debunking the singularity.
Re: "There’s a scenario you don’t seem to be considering, which is a gradual, world-wide development of deliberately designed AIs, leading eventually (over the very long term) to a “singularity”, a lifestyle beyond which it is difficult to make predictions."
Agreed, except: if a “singularity”, is "a lifestyle beyond which it is difficult to make predictions" - haven't we had some of those already? 2000 was surely 1800's "singularity" on those grounds. Since such terminology is so vague and silly, why perpetuate its use?