TL8 Computing

[Anthony]

Originally Posted by AlexanderHowl

Considering that we have several thousand computers that are the equivalent of Fast Megacomputers by GURPS standards, we should have an Emergent AI by now if it was possible with our current technology.

It is unlikely that emergent AI can emerge on our computers, but that has nothing to do with computer power; our computers just aren't designed in a way that will let it happen.

[Fred Brackin]

Originally Posted by AlexanderHowl

Complexity is defined as a 10x increase in capabilities for each +1 increase, which is a quantifiable measure. I am suggesting that the progression of computing in TL8 was greater than suggested. My parent's old 286 is an early TL8 computer while my laptop is a late TL8 computer, and the difference is over 10,000x (memory, storage, transistors, etc), suggesting a +4 increase in Complexity since GURPS gave us a quantifiable measure.

You are (in my judgment) misunderstanding "capability" to be the same as "numerical measurement of technical capacity".

I'm with Rupert in saying that my computer of 2018 isn't even 10x as useful as the first one I had in 1998.

[Johnny1A.2]

Originally Posted by Icelander

I strongly suspect that any future TL system will not put the 1980s and 1990s at the same TL as the 2000s and 2010s.

Too many basic assumptions about how society works, technology is used or how PCs use their skills are different. A case might be made that the ubiquity of computers, personal comminication devices, the Internet and constant data connection has already caused the biggest change to societies in recorded history.

Compared to the social transitions from nomadism to agriculture, the mechanization of agriculture (which moved us from a society in which over 50% of every society worked in agriculture one way or another to one where it falls toward 1%, and the resulting mass urbanization), or the impact of instantaneous communications in the 19C, or the internal combustion engine? Heck, compared to the car?

Computers and IT seem like a big deal because they are recent, and because, as Robert Heinlein noted, 'a nine days wonder is taken for granted on the tenth day', so the bigger changes over the last few centuries are taken as part of the background. But they're not even in the same league, so far, as those earlier changes.

[Johnny1A.2]

Originally Posted by Ashtagon

I tend to agree with David. While it's true that the raw disk space available has grown explosively, there is a law in computer science that the operating system will grow to fill the disk space available. My Windows 10 machine migt have prettier graphics, but for the end-user tasks (as opposed to the OS-generated tasks) assigned to it, it isn't literally 1000 times faster than my first PC (a 486 25 mhz, 1 meg ram, with a 20 meg hard drive running DOS 6.22 and windows 3.11 when I could be bothered to boot into that memory hog).

Relevant: http://www.totalgeekdom.com/?p=356

Someone once asked Steve Wozniak when software would be as reliable and efficient as hardware, and he replied to the effect that it would happen when hardware improvement leveled off, removing the pressure toward feature bloat and encouraging efficiency.

As long as the hardware keeps getting more powerful, the software engineers have no strong motive to make the software more efficient, and pressure from marketing and related areas produces more bloat. Modern web pages download the equivalent of an entire book when sending a page no bigger than the equivalent page from 2000.

[Johnny1A.2]

Originally Posted by Anthony

Yes? While the brain isn't structured much like a modern computer, its raw computing power does not appear to be all that high.

We don't have any idea what the brain's raw computing power is. It's not clear that it's even a meaningful measurement in that context.

[Anthony]

Originally Posted by Johnny1A.2

We don't have any idea what the brain's raw computing power is. It's not clear that it's even a meaningful measurement in that context.

We can get estimates of neuron firing rates (total for the brain probably 10^10 - 10^11); how those compare to cpu instructions we don't know exactly, but given the very limited bandwidth of synapses, probably not dramatically more, so a teraflop is likely adequate processing. That's doable today, though the memory layout is much different.

[Daigoro]

Originally Posted by Fred Brackin

You are (in my judgment) misunderstanding "capability" to be the same as "numerical measurement of technical capacity".

I'm with Rupert in saying that my computer of 2018 isn't even 10x as useful as the first one I had in 1998.

I agree on the end-user experience not changing as much. Heck, I've gotten familiar again with the ol' blue screen of death since the recent Windows 10 updates.

But for gaming purposes, the kinds of things you're doing would benefit from the x10,000 speed increase. I'm thinking of things like database searches or brute force password cracking. Picture handling also- I vaguely remember a scene from some action movie where they were painfully waiting for the picture of a suspect to slowly raster onto the screen, whereas these days we can basically replicate Deckard's analysis of a Replicant's family photo.

I'd second the recommendation for the Pyramid Thinking Machines article, it goes a long way to fixing some issues raised (although it has its own problems- such as multiple types of computer at the same TL having the same Complexity rating).

[Say, it isn't that bad!]

If you want a morbidly hilarious (and technical) look at just how badly some stuff is programmed, this blog is excellent:

https://rachelbythebay.com/w/

[Celti]

As Flyndaran and Daigoro suggested, if you want a really comprehensive treatment of this sort of thing, snag a copy of Pyramid #3/37: Tech and Toys II and look at the article Thinking Machines. It handles computers in Complexity levels matched roughly to flops, multiplying by 1,000× for each Complexity level. The TL progression is divided into quarter-TLs stretching from 6.75 to 10.75, matched fairly well to dates between 1930 and today. At TL11 and 12, Complexity is based on molecular computing with an assumed efficiency of bits per atom.

Its two main flaws are silence on the subject of storage costs — which is understandable — and the fact that TM Complexity does not line up precisely with RAW Complexity. For the latter, I put together a rough correspondence and a list of software and costs at that Complexity:

• Old: 1-2 New: 3 Cost: $50 Software: Software tools (basic); target tracking (basic); virtual reality (basic); word processor (good).
• Old: 3-5 New: 4 Cost: $500 Software: Cosmetic filter, holotech editor (basic); software tools (good); TacNet (good); targeting software (improved); translator (one-to-one); virtual reality (full); visual enhancement; word processor (fine).
• Old: 6-8 New: 5 Cost: $5,000 Software: Dream teachers; dreamgames; holotech editor (improved); sensie editor (basic); sensie player/uplink; software tools (fine); TacNet (fine); TacNet server (basic); thought processor; translator (universal); virtual reality (total); word processor (best).
• Old: 9-11 New: 6 Cost: $50,000 Software: Sensie editor (improved); software tools (best); TacNet server (best).

[Icelander]

Originally Posted by Johnny1A.2

Compared to the social transitions from nomadism to agriculture, the mechanization of agriculture (which moved us from a society in which over 50% of every society worked in agriculture one way or another to one where it falls toward 1%, and the resulting mass urbanization), or the impact of instantaneous communications in the 19C, or the internal combustion engine? Heck, compared to the car?

Computers and IT seem like a big deal because they are recent, and because, as Robert Heinlein noted, 'a nine days wonder is taken for granted on the tenth day', so the bigger changes over the last few centuries are taken as part of the background. But they're not even in the same league, so far, as those earlier changes.

You're arguing that the combined effects of ten thousand years and seven TLs are more than the shift from 1 TL. I don't disagree.

The mechanisation of acriculture wasn't one TL of change. Nor was urbanisation. Urbanisation has increased in proportion with TL and the percentage of people who worked in acriculture has fallen steadily since TL3.

The most industrialised societies fell below 50% acricultural workers at late TL5, but some societies haven't reached that yet. There was no one TL shift that took us from an agrarian society to a post-industrial one. Some societies went from effective TL5 to effective TL8 in living memory, but that change, while rapid, isn't a one TL shift.

Note also that cheap cell phones and the much less expensive infrastructure required compared to land lines mean that TL8 has brought instananeous communication to huge parts of the world and significant fractions of humanity that have effectively never had access to any such thing.

Not to mention that having daily social contact with people from many different cultures, who live in different parts of the world, has shifted from being a feature for the super-rich elite, who travelled the world easily and could ignore the high costs of transcontinental cables and international phone calls, to being merely for the affluent on the scale of the world as a whole (i.e. anyone in Western countries and many other, richer societies, the luckier in many others), those who can afford a computer/smartphone and connection.