Transhuman Space in Infinite Worlds

[jbalsle]

Originally Posted by johndallman

Consider your reaction to this argument. If it's meaningful, you're tending to the THS(Hard) style, if not, then not.

Oddly enough, my response is 'well, what if he wants it to take weeks to beam an AI? Shouldn't it be his choice as GM?' Where does that fit in this 'hard' vs. 'soft' debate? :)

I already stated that I preferred beaming of AI from Earth to Mars to be as common as in THS, because that opens up interesting avenues for GMing and playing a game. But in a game where the GM is trying to stop that? It should be his right to choose his mechanics, and my right to choose to play or not.

I guess you could call my choice of hard vs. soft to be a soft choice, rather than a choice to be hard or soft....

[jbalsle]

Originally Posted by vicky_molokh

This is a case of Has No Sense of Proportion. Error correction doesn't work like that (in the real world).

You know this. So, incidentally, do I. But perhaps in his game world, it does, or perhaps there's something special about AI -- maybe it's Super-Turing or something similar, that means if you beam AI, you risk it not arriving at its destination in one piece. You may not like it, and neither may I, but hey, if you don't, there are ... err, at least a few other THS games you can play in. But let's not nitpick the heck out of this guy's game saying how impossible it would be. He wants his games to be soft. Let him have what he wants.

Changes like the one mentioned have consequences. ...trying to enforce all AIs (and Ghosts) to be hardware robots will break (read: change) the setting in ways that take tens or hundreds of pages to analyse and summarise.

And this is the crux of what I was trying to say, just without the use of the loaded word 'break'. It's his game, tis all I was saying. Lets help him with that analysis and summarization, and help him build something that can foster a Sense of Disbelief in players so he can go about doing the main thing we all come to this hobby for -- having fun.

[Flyndaran]

I just don't see how the ability to send sapient A.I.s digitally at light speed is that fundamental to THS. It's definitely not among the first 10 things I think of when thinking of THS.

[johndallman]

Originally Posted by jbalsle

Oddly enough, my response is 'well, what if he wants it to take weeks to beam an AI? Shouldn't it be his choice as GM?' Where does that fit in this 'hard' vs. 'soft' debate? :)

The difference is in how you achieve the result. If you're doing it "soft" you just say it's impossible. The difficulty with that is that if you make several rulings like that and they're not consistent with each other, the world can stop making sense to your players.

If you're doing it "hard", you need an explanation of why it's impossible, or impractical, and you need to consider what else that affects, and how that changes the setting. This is more work, but means that the result is internally consistent. That means that when the players want to do something creative, you have strong grounds for deciding how difficult it is.

For this particular case, a "hard" justification for it being impossible to transmit AIs is tricky to achieve. Impractical is easier.

The figures for how fast you can transmit data by laser are reasonably firmly based in simple physics and information science. To use an analogy, saying that it takes a couple of weeks to transmit 100TB in 2100 is like saying that it will take a couple of weeks minimum to travel from New York to San Francisco in 2100. It doesn't seem plausible, because we can do better than that now, and there's nothing about the setting that indicates that travel will get slower.

Another way to restrict travel-by-laser might be to make smart AIs much larger. If that Complexity 8 AI is 1000 times larger, 100PB rather than 100TB, it takes 1000 times as long to transmit, which takes 2-3 months. This is a much less attractive way to travel, so we're looking good so far. However, the much larger AI still has to be able to run on a computer, and your computer needs 1000 times as much storage to hold it. That means that either storage needs to have 1000x the capacity, which is going to have effects that I'd need some time to work out, or the computers that run AIs need to get a lot bigger and more expensive.

That breaks something that is a THS trope: that AIs can run on computers that fit into human-sized bodies. You can salvage that trope by changing the size progression for program complexity to something like this: no change up to Complexity 4 (still 10GB, the largest programs we have today) then Complexity 5 is 1TB, Complexity 6 is 100TB, Complexity 7 is 10PB, Complexity 8 is 1000PB, and so on getting 100 times larger for each +1 complexity. That allows not-very-smart AIs to be mobile, but smart ones need huge computers. How does that seem?

[Flyndaran]

I like my idea for ground up A.I.s and why transmitting them would cause major problems.
It allows the transmission of completely fresh out of the box A.I.s which might be a good or bad thing depending on how the GM wants the setting to progress.

I can't think of a reasonable way to give ghosts the same limitation.

[The Benj]

If you want non-transmissable sapience, massive storage overheads and massive storage seems like the way to go, yeah.

[Flyndaran]

Originally Posted by The Benj

If you want non-transmissable sapience, massive storage overheads and massive storage seems like the way to go, yeah.

That would have the secondary effect of making computers housing such A.I.s much larger.

[The Benj]

Originally Posted by Flyndaran

That would have the secondary effect of making computers housing such A.I.s much larger.

No, by "massive", I mean highly capacious. Increase the amount that storage can hold in the setting, but not the amount of bandwidth.

[Fred Brackin]

Originally Posted by johndallman

If you're doing it "hard", you need an explanation of why it's impossible, or impractical,?

For example, you could say that AIs can't self-install themselves with the ease of a JavaScript update. I know several people who have very well paying jobs just keeping databases smaller than a TS AI in working order and databeses are much simpler than a full AI will be..

So maybe you can beam AIs but it's no good without someone at the receiving end to oversee the installation into hardware.

[sir_pudding]

I'm a little confused here, anyway. If an SDV is transported to Steel, then where are they going to be sending their infomorphs to exactly? I'd think transmitting an infomorph to a zonemimd would be a really bad idea, essentially giving them an editable POW.