Complexity for computers and the real world

[Anthony]

Originally Posted by maximara

The problem here is you are comparing two totally different instruction sets RISC and CISC.

No, the problem is mostly that clock speed comparisons haven't been particularly useful for anything for 15-20 years, and even before that weren't terribly relevant between different architectures. In any case, the main feature of the M1 is that it has a very low power draw for its computing power.

[Kage2020]

So. How does this getting translated to something fun for people that aren't nerding out on processor speeds etc.?

Don't get me wrong, I've done the same myself in the past when it comes down to how archaeology is represented in GURPS. I think that it's easy to do when you have a specialism in something and you want to get more love or "realism" in the game.

With that said, what this thread has illustrated is that there should be an expansion in the computer rules in GURPS that don't try to mimic the real world but try to mimic the facets of computers that would impact across the spectrum. (Ala, Kromm.)

Classic Cyberpunk, for example, "Standard Computer Speed" that was linked to Complexity (p.74). Even this introduction offers a multiplication of options in representing computers if you want to explore that in your setting.

What are the abstractions that would be meaningful in GURPS as they applied across the genre?

[Kromm]

Originally Posted by Kage2020

What are the abstractions that would be meaningful in GURPS as they applied across the genre?

Some rating of how many programs you can run at once ("capacity"). Very important . . . vigilante forensic scientist running database searches on a "personal project" in the background while running analyses in the foreground, starship computers that can navigate and target bad guys at the same time, phones that can take calls while displaying an assisted GPS map for navigation, etc. This is probably "good enough" to fake number of users, but in some campaigns you might need to specify that; e.g., a computer that can run 20 programs but has poor or no virtualization isn't the same as one that can spread out the programs it can run across up to 20 sessions, each of which might be hard-capped to use limited resources, so even if there's just one user, they can run only one program, not 20.

Some rating of speed. This would add a bonus or remove a penalty, mostly. It might also reduce execution time enough to affect adventures; e.g., protein-folding related to your miracle vaccine research in a day vs. a week vs. a month makes a huge difference. You could tie the two together: A computer that's much faster than the minimum can effectively give a bonus for taking extra time. This isn't always linked directly to capacity – for real computers, it isn't generally true that if you run one program and not 20, that one program runs 20 times as fast. Not all code is "vectored" or "multi-cored" or "multi-threaded"; not all machines support code that is. You'd need a way to flag that, too.

Some rating of how helpful a program can be run. This determines the quality bonus that the software and hence the computer gives to any task that requires a computer or just benefits from one. This needn't be linked to either of the above. It's often less a function of capacity or speed than it is a function of things like graphics quality, sound quality, the type and quality of human-interface devices supported, etc. Also, the machine's architecture could be optimized for specific kinds of code; two computers of equal capacity and speed could be tweaked for targeting vs. navigation, 3D rendering vs. machine learning, etc., etc., etc.

I'd say all of those are fairly universal in importance. To this, add things that aren't strictly about the computer, like mass storage and network speed. It's easy to say "buy them separately," but if you're trying to be realistic, there will be limits on "how much," "how fast," and so on set by the computer itself. A huge advantage of modern PCs is the vast quantities of data they can access locally or over a network.

BUT . . . I think it would be important to avoid hard associations between these things and real-world measures like central processor speed, number of cores, number of threads, mainboard architecture, bus speeds, quantity of memory, memory speed, type of storage interface, graphics processor speed, number of graphics cores, etc., etc., etc. That stuff is too "world specific" to be generic, and is driven as much by marketing, competition, patent protection, and the like as by any hard limits on technology. Also, all the possible interactions between these things and specific software are next-to-impossible to predict.

[AlexanderHowl]

I think that Complexity is best understood as capabilities relative to other computers within the same TL. Given the history of advances from 1981-2021, I would categorize computers into Very Early (-2), Early (-1), Mature (+0), Late (+1), and Very Late (+2) categories within the same TL, modifying the Complexity as above. When comparing TL(X) to TL(X+1), I would add +6 Complexity.

Within the same TL, base Complexity for Mature computers would be also modified by TL as standard, so a Mature TL7 desktop has Complexity 1 compared to other TL7 computers while a Mature TL8 desktop has Complexity 3 compared to other TL8 computers. I would dispense with the Complexity modifiers for vacuum tubes, transisters, etc., as they would be included by default in the Very Early to Very Late categories. The other Complexity modifiers could remain though, so you could have Slow or Fast computers.

In that case, the laptop computers of 2021 would be Very Late TL8, giving them an effective Complexity 5. When compared to the laptops of 1981, they have 10,000x the capabilities, which sounds about right. When compared to Very Late TL7 desktop computers, they would have an effective +6 Complexity, which also sounds about right. Relative storage would increase by 10x per +1 Complexity, which also sounds about right, as storage gets very cheap.

[maximara]

Originally Posted by Kromm

Some rating of how many programs you can run at once ("capacity"). Very important . . . vigilante forensic scientist running database searches on a "personal project" in the background while running analyses in the foreground, starship computers that can navigate and target bad guys at the same time, phones that can take calls while displaying an assisted GPS map for navigation, etc. This is probably "good enough" to fake number of users, but in some campaigns you might need to specify that; e.g., a computer that can run 20 programs but has poor or no virtualization isn't the same as one that can spread out the programs it can run across up to 20 sessions, each of which might be hard-capped to use limited resources, so even if there's just one user, they can run only one program, not 20.

Funny things about that one. Cloud computing messes with it - a lot. Of course in reality the program isn't being "run" on the computer but you still need communication speeds and processing power (for the graphics) to do this.

Might want to include an "overhead" modifier - a GPS that only displays text compared to one that is also able to generate a 3D image of area you are in.

It is the TL8-9 version of the old TL7 dumb terminal connected to mammoth server set up.

Originally Posted by Kromm

Some rating of speed. This would add a bonus or remove a penalty, mostly. It might also reduce execution time enough to affect adventures; e.g., protein-folding related to your miracle vaccine research in a day vs. a week vs. a month makes a huge difference. You could tie the two together: A computer that's much faster than the minimum can effectively give a bonus for taking extra time. This isn't always linked directly to capacity – for real computers, it isn't generally true that if you run one program and not 20, that one program runs 20 times as fast. Not all code is "vectored" or "multi-cored" or "multi-threaded"; not all machines support code that is. You'd need a way to flag that, too.

Basically the scene in New Hope where Han tells Luke that they need to have the computer calculate their path as not doing this results in really bad things using hyper jump - like flying into a sun ala Yahtzee's third character in Elite: Dangerous

This too would need some form of modifier - running software that is very new on an old computer (that sin't expandable for design or monetary reasons) tends to be much slower than running software made for it when it was brand new.

Think of the Cyberpunk netter wholesaled can't afford the latest datalink hardware and has do with some antiquated hardware from 10 years ago but also has to run as new as possible datalink software on the thing.

Originally Posted by Kromm

Some rating of how helpful a program can be run. This determines the quality bonus that the software and hence the computer gives to any task that requires a computer or just benefits from one.

Classic tried to do something like this with complexity applied to program type

Datalink: 1 vs Personality Simulation: 5.

The one problem I have with both the old and new system is the cost factor.

Even if you don't adjust for inflation hardware tends to get cheaper even as with the M1 Macs with what have what appears to be "cutting-edge technology" (x20 cost) that is out performing any 10 Watt entry level machine you can name and yet it is cheaper then the equivalent Macs of just a year ago.

[Kromm]

Yeah, if modeling the real world, capability seems to go up even as price goes down. A lot of that has to do with economics, not tech per se:

When I was a kid in the late 1960s and early 1970s, nobody much had computers of any kind at home unless they were electronics engineers or eccentric millionaires. When I was in my teens in the 1980s, enthusiasts (a.k.a. nerds) and people with rich bosses (and the rich bosses themselves!) had desktop computers at home and perhaps clunky portables you couldn't reasonably call "laptops." Through the 1990s, people slowly warmed to desktop computers, and laptops became viable for people of modest means; many people had one or the other. Since then, people can't imagine not having computers, often one desktop or laptop per person in the household, sometimes one of each for certain people, plus tablets, smartphones, gaming consoles, . . . just in this room, we have a desktop, two laptops, a tablet, two smartphones, and a smart TV.

So the tech was improving, sure! But more important, the market was growing at an astronomical pace. Specialist and professional markets expanded to many times their original size as enthusiasts appeared, "enthusiasts" came to mean "the entire middle class," that quickly turned into "almost everyone who isn't homeless, and even some of them" – and on top of that, the count went from "one per person in the target market" to "several per person in the target market." That led to massive economies of scale, vicious consumer-market competition driving prices down and at times creating loss-leaders, planned obsolescence bringing in more dollars, and the whole thing sucking in so much consumer money to fund R&D that the tech improved even faster and the whole cycle accelerated.

That isn't generic, though. There are tons of assumptions there about markets and consumer behavior and Zeitgeist that might not repeat themselves even if we reran our history. In a fictional world, any small change could lead to a "better computers cost more and more over time" model. So I think that in a generic game, it might be best to peg price to capability at each TL, and then write a little worldbuilding box offering alternative pricing for different economies . . .

[Kromm]

For instance, in a high-Control Rating command economy in some backwater insulated from globalization (or its interstallar or intergalactic or crossworld equivalent), El Jefe can just declare that computers cost $x per MHz or MB or FLOPS or whatever, using the state-owned computer industry to fund the imperial war machine while deliberately keeping the best stuff out of citizens' hands. Whatchagonnadoaboutit?

Whereas in some post-scarcity setting that has embraced human obsolescence, computers might bootstrap themselves to ever-better levels in real time, and nobody really pays anything at all.

Even being less silly, I suspect that if we hadn't had the economic boost of post-Depression recovery and the WWII war effort, with a war against Nazis won more than in part thanks to clever tech (sonar, radar, etc.), we probably wouldn't have had the post-war love affair with "the Space Age" and all things techie, nor the investors to turn that into consumer goods. It's entirely possible we'd still be stuck with so-so electromechanical mainframes.

[David Johansen]

Originally Posted by Kromm

BUT . . . I think it would be important to avoid hard associations between these things and real-world measures like central processor speed, number of cores, number of threads, mainboard architecture, bus speeds, quantity of memory, memory speed, type of storage interface, graphics processor speed, number of graphics cores, etc., etc., etc. That stuff is too "world specific" to be generic, and is driven as much by marketing, competition, patent protection, and the like as by any hard limits on technology. Also, all the possible interactions between these things and specific software are next-to-impossible to predict.

Reading old rpg computer specifications in games like Rolemaster Blackops, The Morrow Project, and Heroes Unlimited is always amusing.

[maximara]

Originally Posted by Kromm

That isn't generic, though. There are tons of assumptions there about markets and consumer behavior and Zeitgeist that might not repeat themselves even if we reran our history. In a fictional world, any small change could lead to a "better computers cost more and more over time" model. So I think that in a generic game, it might be best to peg price to capability at each TL, and then write a little worldbuilding box offering alternative pricing for different economies . . .

Originally Posted by Kromm

For instance, in a high-Control Rating command economy in some backwater insulated from globalization (or its interstallar or intergalactic or crossworld equivalent), El Jefe can just declare that computers cost $x per MHz or MB or FLOPS or whatever, using the state-owned computer industry to fund the imperial war machine while deliberately keeping the best stuff out of citizens' hands. Whatchagonnadoaboutit?

The problem with that example is it is not generic either. One can just as easily argue 'In a fictional world, any small change could lead to a "better computers cost less and less over time than in our world" model.'

My fanfiction Princess Morganite is one such world where the change was back 12,000 years ago with Beryl getting a fraternal twin sister. Using fragmentary Clarktech of the long dead Earth Kingdom Beryl has pushed computer development to where their 1990s' computers are on par with our 2020 ones:

"Good. I setting you up with Mercury. See if she can pull anything out of her computer we can reverse engineer. There is the potential of being able to make stuff that will kick the current 64-bit 5 watt HT100's to the curb."

"That will make your favorites happy. We have been basically cranking out that CPU for the last five years with next to zilch improvements."

"It's taken them this long to access some of the chip's true power. Still working on the 1 TB RAM design I see," commented Beryl as she went over the material on the clipboard. "There is such a thing as too far ahead of curve. We aren't trying to recreate the technology of the Earth Kingdom overnight. Only fast enough that factories Earthside can make what we come up with. Spend more time over in software design. Let's make what we have now shine," finished Beryl handing the clipboard back.

---
In canon computers are better in that world than our own in that period (1992-1995) though not on par with the Magitech computer Mercury has.

Another anime that has software ahead of ours is Swordart Online and I doubt we will get to that level of VR by 2022.

The Axis of Time has computers both military and commercial ahead of ours in 2021 (They had Hilery Clinton as president) and they are causing technology to progress much faster than in the alternate timeline that they have created.

Mark Twains 1898 "From The ‘London Times’ in 1904" which describes the telelectroscope the foundation for the internet like system in the story.

As I said before computers have been one of the things that we have underestimated the development in most of our fiction.

The thing it is seams that most technological development is a Great Moment rather than a Great Person (The whole premise of Burke's 1970s Connections series) and the computer is one such device.

[whswhs]

Originally Posted by maximara

The problem with that example is it is not generic either. One can just as easily argue 'In a fictional world, any small change could lead to a "better computers cost less and less over time than in our world" model.'

Well, yes, but the point of the example seems to be precisely to establish that different histories could have different rates of progress. Your fictional world is simply another case of the same principle. It may be less striking, in that it illustrates "fast progress," which our world also illustrates for many people; Kromm's hypothetical world illustrates "slow progress." But both illustrate the fundamental point of diversity.