Notes on the Singularity.
Jul. 12th, 2004 06:01 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
swestrupI was fooling around with Mathematica today, and after an large amount of frustration, I managed to get it to spit out this chart:
Its a simple thing, but I'm quite proud of it. It doesn't have the axes labelled or anything, because I was too lazy to figure out how, but it has the shape I was looking for.
This curve represents cumulative technological progress in the next century, correcting for 'inflation' in the rate of progress. So, the X-axis represents actual years from 2000 and the Y-Axis represents cumulative years of technological progress, measured in multiples of the amount of progress we made during the year 2000.
Now, the first thing you may notice about this chart, is that its not linear. Nope. The rate of technological change is increasing and this chart reflects that. The second thing you may notice is that the number at the top of the Y-Axis is 20,000. As in, years. So, this chart predicts that, assuming that the rate at which technological change was increasing during 2000 is constant, during the next 100 years we will see 20,000 years-equivalent of progress.
You should keep this in mind whenever almost any scientist says that some technology or other is 'three to four hundred years in the future'. When they say that, they are assuming the rate of technological change is constant, when it isn't. This is to help translate such estimates into real-world numbers. So, pulling out some points from that chart and sticking them in a table we see:
They Say It Takes
-------- --------
50 yr 21.07 yr
100 yr 29.03 yr
500 yr 49.72 yr
1000 yr 59.05 yr
5000 yr 81.00 yr
10000 yr 90.49 yr
So, that estimate of 300-400 years can be expected to happen before 2050.
This would all be hard enough to swallow, but the rate of change of technological progress during the year 2000 wasn't constant. It was increasing. What this means is that this chart is a conservative estimate. We don't really have a very good handle on the rate at which progress is accellerating, nor do we know that the increase is merely hyper-exponential. It may be MUCH faster. When people talk about the 'Singularity', they are usually postulating that sometime during the next hundred years that line above goes to infinity, not just 20,000.
What would that mean? No one is exactly sure, but it certainly is fun to speculate... Of course, its coming fast enough that the majority of folks on LJ will live to see it, for good or bad.
Its a simple thing, but I'm quite proud of it. It doesn't have the axes labelled or anything, because I was too lazy to figure out how, but it has the shape I was looking for.
This curve represents cumulative technological progress in the next century, correcting for 'inflation' in the rate of progress. So, the X-axis represents actual years from 2000 and the Y-Axis represents cumulative years of technological progress, measured in multiples of the amount of progress we made during the year 2000.
Now, the first thing you may notice about this chart, is that its not linear. Nope. The rate of technological change is increasing and this chart reflects that. The second thing you may notice is that the number at the top of the Y-Axis is 20,000. As in, years. So, this chart predicts that, assuming that the rate at which technological change was increasing during 2000 is constant, during the next 100 years we will see 20,000 years-equivalent of progress.
You should keep this in mind whenever almost any scientist says that some technology or other is 'three to four hundred years in the future'. When they say that, they are assuming the rate of technological change is constant, when it isn't. This is to help translate such estimates into real-world numbers. So, pulling out some points from that chart and sticking them in a table we see:
They Say It Takes
-------- --------
50 yr 21.07 yr
100 yr 29.03 yr
500 yr 49.72 yr
1000 yr 59.05 yr
5000 yr 81.00 yr
10000 yr 90.49 yr
So, that estimate of 300-400 years can be expected to happen before 2050.
This would all be hard enough to swallow, but the rate of change of technological progress during the year 2000 wasn't constant. It was increasing. What this means is that this chart is a conservative estimate. We don't really have a very good handle on the rate at which progress is accellerating, nor do we know that the increase is merely hyper-exponential. It may be MUCH faster. When people talk about the 'Singularity', they are usually postulating that sometime during the next hundred years that line above goes to infinity, not just 20,000.
What would that mean? No one is exactly sure, but it certainly is fun to speculate... Of course, its coming fast enough that the majority of folks on LJ will live to see it, for good or bad.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2004-07-12 04:10 am (UTC)no subject
Date: 2004-07-12 05:27 am (UTC)Secondly, it's contrary to experience. When we're told something will happen in a couple of years it typically seems to take twenty or thirty... Well, not always, I admit, but generally one is astounded at how slowly things move....
no subject
Date: 2004-07-12 05:39 am (UTC)(b) That's the spread between 'can do' and 'did'. But the big change that's happening at the moment is that first communication technology and now fabrication technology is being, um, give-to-the-people-ised (for which there's a word, but I'm still half asleep).
Yes, the poers that be are fighting it. It's why the stink about european software patents, for example: that's an attempt to make it difficult for private citizens to write software. As if that cat is going back into the bag.
no subject
Date: 2004-07-12 05:43 am (UTC)We'll see if his strategy works.
But in Europe, children use their cellphones to go, see what I'm doing?
But in Korea, cellphones have the bandwidth that US televisions do.
Basis of calculations
Date: 2004-07-12 11:45 am (UTC)Would you happen to have the source reference information for your calculations hanging about? Something this....novel needs a second or more set of eyes to interpret the data - both forward and back (wouldn't want to base a forecast on one anomalous year...)
Ross
no subject
Date: 2004-07-12 07:51 pm (UTC)2) Okay, I actually confess to a minor flaw in the math on this point. People actually make TWO errors in prognostication. The other is that things are usually harder to do than people think they are, thus adding a linear factor to the results. The trouble is, I have no information about the value of this factor. I suppose if one did a study of project time overruns for short run projects (1-5 years) where the rate of technological progress is slow enough may be assumed linear over that period (ie, again not in IT), one could estimate it. To my mind it could be anywhere from 1 to 10, although I would assume its less than 2. I seriously thought about multiplying all estimates by 3 before looking it up on the curve. That would mean that a 5 year estimate would end up taking 10.15 years, and a 300-500 year estimate would take 58 to 65 years to complete rather than the 43 to 50 years that my chart gives. The only reason I didn't do it was I didn't think about it until half way through my post, and I was getting very sleepy.
Re: Basis of calculations
Date: 2004-07-12 08:44 pm (UTC)I can probably dig up MANY more references later, if you're still interested.