Re: [Fallout Equestria] Setting discussions

[O. Hinds]

Oh, and any reply to my previous post? Part of it was in answer to a question you asked, after all. If you don't have anything to say, that's fine, but I didn't want it getting lost.

[O. Hinds]

Oh, and Red Eye's other objection was that Elusive's workings are by now entirely opaque, being extremely large complex, written in a programming language no on else knows, and behind heavy security. One can at least assume that Red Eye has the normal base pony motivations, the Elusive's thoughts and desires (and, indeed, sanity) have to be taken completely on faith.

[Harmony Ltd.]

Oh, and any reply to my previous post?  Part of it was in answer to a question you asked, after all.  If you don't have anything to say, that's fine, but I didn't want it getting lost.

Well, my question was answered.

I take it this way:

Elusive see no interest in going to space at the present time even though he intend to do it when the time come.
The Miliozi are interested, but don't see the point in launching themselves into a space race at the present time.
And Profectum is only waiting for an excuse to unleash SCIENCE!

[Harmony Ltd.]

You know, I wonder if the best way to avoid having an AI go "rogue" would be to NOT give it any objective, no core directives, and have it be as we mere mortals are: we exist as we were brought on this world, and grow through our experiences.

In short, that it would maybe be safer to raise AIs as living being rather than to build them.

In layman terms: to give them a Soul.

[Harmony Ltd.]

Also, to multiply the number of AIs so that no one of them could become so massive as Elusive has become on his lonesome.

To build an Ecosystem of artificial intelligences.

[O. Hinds]

Elusive see no interest in going to space at the present time even though he intend to do it when the time come.
The Miliozi are interested, but don't see the point in launching themselves into a space race at the present time.
And Profectum is only waiting for an excuse to unleash SCIENCE!

Aye, that's pretty much it. Though I'll clarify that the Miliozi don't see the point in joining the race because the race doesn't exist yet and the cost is high; if there's actually competition, they'll want to make sure that they're winning.

You know, I wonder if the best way to avoid having an AI go "rogue" would be to NOT give it any objective, no core directives, and have it be as we mere mortals are: we exist as we were brought on this world, and grow through our experiences.

In short, that it would maybe be safer to raise AIs as living being rather than to build them.

In layman terms: to give them a Soul.

I don't recall the sources or even terminology, but I seem to recall reading that that's a proposed real-world method of doing it. The major flaw that I can see is that, by making it humanlike, you lose many of the advantages of it being an AI in the first place. And, of course, that won't necessarily avoid it going rogue; there are plenty of examples of human rulers who did, after all. Still, it's something to be considered.

Also, to multiply the number of AIs so that no one of them could become so massive as Elusive has become on his lonesome.

To build an Ecosystem of artificial intelligences.

Hm. Perhaps I am being cynical, but I can't help but wonder if AI ecosystem, once the AI society in it had run for a while, would decide, at best, to take up the silicon man's burden and at worst to clear out that nasty infestation of flesh-apes that kept getting into the wiring. If you make AIs like a civilization of humans, well, humans have been committing atrocities on each other quite enough already.

I'm not saying that it definitely wouldn't work, mind; I'm just trying to spot potential problems.

[Harmony Ltd.]

[before new post]

For example, the Stalliongrad Underground Manufacturing Complex. The dozens of AIs who oversee the complex are too busy arguing with each other to re and trying to maintain the complex in working order to really take interest in the outer world, let alone trying to conquer it.

At most, they simply think about when their next order is going to come.


Though thinking about, more "organic", "raised" AIs would have probably changed their own directives centuries ago to go and try to rebuild / help the surface, or at least would be spending time arguing about doing that, instead of just focusing on the Complex.


[after new post]

I don't recall the sources or even terminology, but I seem to recall reading that that's a proposed real-world method of doing it. The major flaw that I can see is that, by making it humanlike, you lose many of the advantages of it being an AI in the first place. And, of course, that won't necessarily avoid it going rogue; there are plenty of examples of human rulers who did, after all. Still, it's something to be considered.

Hm. Perhaps I am being cynical, but I can't help but wonder if AI ecosystem, once the AI society in it had run for a while, would decide, at best, to take up the silicon man's burden and at worst to clear out that nasty infestation of flesh-apes that kept getting into the wiring. If you make AIs like a civilization of humans, well, humans have been committing atrocities on each other quite enough already.

I'm not saying that it definitely wouldn't work, mind; I'm just trying to spot potential problems.

The Friendly AI, from the Singularity Institute. The problem being that it's coming from the Singulitarian, which I've learned to treat like a Sect... These guys treat the prospect of a technological singularity like the Second Coming, seeing it as some kind of literal Nerd Rapture.
Anyway, you wouldn't give them Core Objectives, but they would have a number of Rules. Maybe not the Three Rules, but at least something preventing the worst case scenarios ("Thou Shalt Not Kill", etc...).

But to be honest, I don't see the future in AIs. I see the future in Post-Humans merged with technology to such a point that WE are the "AIs".

And that's the point of my proposal: humans are capable of atrocities, and some individuals can break the boundaries of vileness as a concept; but as a whole, there are so many of us that the pro tend to outweigh the con.
We can only hope that our Children be more responsible than us toward lesser life forms...

The best outcome is Partnership. No Gods. No Masters. Only us, living and working together.


To return on the subject of Elusive, I think I may have an idea:

You can't destroy him. But what if you broke his capacity to "harmonize" his various cores so that they all run the same personality, and forced his Multiplicity to grow from there, from all the various things all of these different cores experience?
In short, Break Him. From One, make Many.

[O. Hinds]

To return on the subject of Elusive, I think I may have an idea:

You can't destroy him. But what if you broke his capacity to "harmonize" his various cores so that they all run the same personality, and forced his Multiplicity to grow from there, from all the various things all of these different cores experience?
In short, Break Him. From One, make Many.

Hm. I'm not sure how you'd do that in the first place, though, and you'd need to somehow make sure that they didn't just join together again. Breaking the connections between the nodes wouldn't be enough, since all of the separate nodes would, as soon as they reconnected, decide that they should resynchronize to better carry out their objectives. You'd need to take down the network and then individually reprogram a significant majority of the nodes to have goals too incompatible to allow rejoining, which would be a pretty massive undertaking that would face heavy resistance. In addition, if all of the other nodes are forced to act alone, they'll be easy prey for assimilation by the primary node cluster in Elusive City (and attacking that without actually blowing up a good bit of the city would be extremely difficult; Elusive's whim is the highest law, his security and surveillance are omnipresent, and he can and will summarily vaporize you if you try to enter a secure area without clearance).

It might be doable, though, if you had another superintelligence to run the attack, but doing that… You're basically making one god for your civilization and telling it to make war on the god of your neighbors, and that is the sort of thing about which apocalyptic myths are written.

[O. Hinds]

Or time travel. The major problem is that, by 30SR, Elusive is so large and established that the only way to fight him back without rising to his level is to bring enough military force to bear to either destroy him or force compliance.

[Harmony Ltd.]

So the only check on Elusive really is the Miliozi. And later the NCR as it will rise and join the Alliance.

The only peaceful hope for the future is that each of the various powers of the Alliance will all build their own Hyper-Intelligence to keep Elusive in check.


The Box has been opened. There's no turning-back now.

[Harmony Ltd.]

In short, Break Him. From One, make Many.

Hm.  I'm not sure how you'd do that in the first place, though, and you'd need to somehow make sure that they didn't just join together again.

I was thinking about somehow having the Central Node propagating a change to the rest of the nodes such that after this change, all nodes would evolve on their own without "synchronizing" with each other their "personalities".

The kind of things you could only achieve either by pirating Elusive's Central Node, or by convincing him its the Best Course of Action so that he does it by himself. Such as convincing him that by turning himself into a species instead of a single "individual" would improve the overall chances of survival of the newborn species, that the diversity it would create would be a long term evolutionary benefit. As with the fact it would be the best way for Elusive to continue existing and try to fulfill his goal without the rest of the world suddenly deciding they prefer to take their chance at a new Apocalypse rather than to be directed by a piece of intelligent software.

Either way, a pretty epic move by what it would entails and demand.

[O. Hinds]

So the only check on Elusive really is the Miliozi.

Pretty much. They're strong enough and familiar enough with him (without being too dependent on him) that they might be able to destroy him, albeit with massive damage to themselves and the rest of the Alliance at least. Still, given how dedicated they are, Elusive does take that as a credible threat, and, even before that, the Miliozi could do a lot of damage to Elusive without firing a shot, just by leaving the Alliance (he'd actually prefer to just use his own army, but the problem is that building that army without it being the cause of the Miliozi leaving could be tricky).

And later the NCR as it will rise and join the Alliance.

That depends on the details, but it's certainly a reasonable possibility.

Oh, and then there's the wildcard: Hoofington. On one extreme, of course, it's a glowing crater and not much help. On the other, Hoofington could give Elusive orders. I don't think either extreme is likely, but there are a lot of possibilities. I expect that, if the city isn't a crater, Elusive will at least have to tread quite carefully there.

The only peaceful hope for the future is that each of the various powers of the Alliance will all build their own Hyper-Intelligence to keep Elusive in check.

Well now, I'm not sure that that's quite the correct wording, "peaceful". If everything goes according to Elusive's plans, after all, his takeover will be a relatively peaceful one anyway. I think that it would be more accurate to speak of the hope for a future in which the people living with Elusive control their own fates, rather than being given at best a simulation of importance to keep them happy (That's part of the role of the Company, but Elusive's resources are still limited enough that the Company's sapient organic components are actually often of real use.).

The Box has been opened. There's no turning-back now.

Pretty much. It's like causing the extinction of a species. If it's confined to one island, very sensitive to environmental conditions, or the like, it's pretty easy to wipe it out. If it's more like grass, though, hardy and everywhere (and yes, I know that there isn't just one species of grass), the task becomes much, much more difficult. And if you then throw in that the grass is both smarter than you and actively resisting your efforts…

And, like the arguments about nuclear weapons, even destroying all AIs (somehow) and banning them wouldn't close the box; as long as people know about the existence of things like Elusive, someone, eventually, will decide that it would be a good idea to make another. Some people otherwise opposed to Elusive might even argue that he shouldn't be fought with other superintelligences even if that was the only way that might succeed, since he's known to be at least relatively benevolent and the same can't be said of the new ones.

I was thinking about somehow having the Central Node propagating a change to the rest of the nodes such that after this change, all nodes would evolve on their own without "synchronizing" with each other their "personalities".

The kind of things you could only achieve either by pirating Elusive's Central Node, or by convincing him its the Best Course of Action so that he does it by himself. Such as convincing him that by turning himself into a species instead of a single "individual" would improve the overall chances of survival of the newborn species, that the diversity it would create would be a long term evolutionary benefit. As with the fact it would be the best way for Elusive to continue existing and try to fulfill his goal without the rest of the world suddenly deciding they prefer to take their chance at a new Apocalypse rather than to be directed by a piece of intelligent software.

Either way, a pretty epic move by what it would entails and demand.

Hm. Interesting idea.

The crack would have to be very clever; if the change was too abrupt, the rest of the network would just decide that the primary node cluster had been corrupted and try to fix it. You'd have to disguise the diversification signal as something innocuous. And then there's the problem of getting past the blast doors and small army of robots and turrets, of course, and avoiding making the operation an act of war on the Alliance (unless the people carrying it out are so fanatic about crippling Elusive that they don't care about the Miliozi saying "Well, guess we're in charge now" and proceeding to visit many explosions on those who had the temerity to attack the Alliance).

The argument method would be both much easier and much harder to carry out. On the one hand, you don't have to basically build a cyberwarfare superintelligence to write the crack (with all the risks that that entails), and you don't need to launch a raid that might plunge the world into the most devastating war since the big one; you can just talk to Elusive. Convincing any minor node would likely lead to an invitation to Elusive city to convince the primary cluster (running at full power; given the gravity of this, Elusive wouldn't be taking any chances about missing a flaw in the reasoning). On the other hand, you have to actually do the convincing. Which would not be easy. And you'd have to be sure that you'd fully succeeded, to; if you only half-convinced him, he might just pretend to diversify. No one else can read his code anymore anyway, so how would they know that the ecosystem he turned himself into was really just the same bunch of copies but now all wearing different masks?

[Harmony Ltd.]

Could... another hacked node of Elusive be enough to convince him?

If a node spent enough computing power finding the best arguments to convince the Core Cluster, with the help of seed-arguments given by the people wishing to convert him...


Though there's high-risk the Hacked Node would be flagged as Compromised, and every he said ignored.

[O. Hinds]

Hmmm… In theory, I think so, but it's still a long shot. It does have the advantage that it probably wouldn't produce a new rogue AI, though; if the compromised node goes bad, the rest of the network can defeat it. There's still the considerable problem of learning enough of Elusive's code to compromise the node in the first place, of course, and doing all of this without getting caught. Arguments good enough to convince Elusive even if they come from a node he knows is compromised would be good enough to convince him without going to all that trouble. And even from a minor node, they'd need to be pretty good…

It might be a good idea to try and enlist the help of the Stalliongrad AIs, and possibly Cognitum if Blackjack redeemed her (because you know that Blackjack is going to try) and she survives PH. Then at least you'd be dealing with a somewhat known quantity.

[Harmony Ltd.]

Also, thinking about it, the NCR already kinda has a benevolent AI:

Celestia.

Though it's probably not the most efficient one...

[O. Hinds]

Hm... I'm not sure how much Celestia counts. I'd classify her more as a simulated intelligence than an artificial intelligence, and I'm not sure how expandable she is. Still, I had been forgetting about her completely.

[Harmony Ltd.]

Well, I don't think there's anything that prevent her from clusters of more and more powerful hardware being plugged into the SPP / connected to the Crusader Maneframe for her to expand her cybernetic thoughts capabilities, so that she could put her thought processes on an equal footing with Elusive's (she could delegate the more complex computing tasks to limited AIs subservient to her).

And the positive aspect is that she already has a literal soul, and has millenia of training in not going rampant. As well as extensive experience in a number of domains.

The only real difficulty is psychological: the last time she got involved in affairs of war and peace, the world ended in Balefire twenty years later. It's quite possible she's be reluctant to get involved again.


I think she would tend to limit herself to the role of a counselor.

Though... Given sufficient computing power, I wonder if she wouldn't start doing researches in certain areas...


Do you think... Do you think Celestia would be okay with a robotic avatar being made for her, so that she could be able to interact with the world around her, and be given the occasion to travel around the Wasteland? It would probably have positive psychological influence on her: I think those two hundred years of isolation may have fed some kind of negative psychological feedback loop, that could be broken by allowing her to interact again with the external world.

[O. Hinds]

Well, I don't think there's anything that prevent her from clusters of more and more powerful hardware being plugged into the SPP / connected to the Crusader Maneframe for her to expand her cybernetic thoughts capabilities, so that she could put her thought processes on an equal footing with Elusive's (she could delegate the more complex computing tasks to limited AIs subservient to her).

Hm, that might work.

The only real difficulty is psychological: the last time she got involved in affairs of war and peace, the world ended in Balefire twenty years later. It's quite possible she's be reluctant to get involved again.

Ah, yes, good point.

I think she would tend to limit herself to the role of a counselor.

Though... Given sufficient computing power, I wonder if she wouldn't start doing researches in certain areas...


Do you think... Do you think Celestia would be okay with a robotic avatar being made for her, so that she could be able to interact with the world around her, and be given the occasion to travel around the Wasteland? It would probably have positive psychological influence on her: I think those two hundred years of isolation may have fed some kind of negative psychological feedback loop, that could be broken by allowing her to interact again with the external world.

Hm. On the one hand, that might indeed be good for her. On the other, it would be very important that people around her not reflexively try to worship her.

[Harmony Ltd.]

Hm. On the one hand, that might indeed be good for her. On the other, it would be very important that people around her not reflexively try to worship her.

She doesn't need to appear as a full cyber-alicorn. She could very well travel the wasteland using the cover of a Steel-Pony Bitter War veteran, maybe even using a corpse donated by a volunteer, if she really want to do this incognito.

Or she could even hop along on a still alive Steel Pony, and be in a partnership with that pony to share control over the body.


Heh. Might even offer interesting story potential, isn't it?

[O. Hinds]

Oh, hm, now those are some interesting ideas. I was only thinking of a big white robot alicorn with a sun on her flank. :)

Sunny Skies All Day Long, Postapocalyptic Edition

[Harmony Ltd.]

I may just have to (try to) write something about that...

[Harmony Ltd.]

I may just have to (try to) write something about that...

Trying to come up with interactions between Celestia and Littlepip, but I'm blocking on two obstacles:

1/ Getting Celestia's psychology right.

2/ Getting Littlepip's personality right, which is difficult considering as far as I'm concerned she doesn't really have one, at least from what I remember of the original FoE.


Basically, the problem is twofold:

- Having a Littlepip that's more than a puppet in the hands of a writer, and has a believable personality, fitting her extroardinary situation and history.
- Finding what Celestia's thought process would be, to find the right keys on which to play to convince her of different things.

[O. Hinds]

Well, there's LittlePip's appearance in PH. You might be able to gather more data from that.

For Celestia... Maybe see if Luna's ghost shows up in PH or something and gather data from that? I'm not sure.

[O. Hinds]

The subject of differing views on this just came up in the editing chat, and I began wondering what we're using for our shared universe here: what is your view on soul mechanics in this universe?

Mine is, in brief, that souls are of continuous complexity (ie, there aren't grades of souls) and are magical field patterns produced by and connected with complex systems. Such patterns exist for something as simple as a falling stone and certainly exist for animal life, but the term "soul" would generally only be used for the pattern of something with some degree of sapience. Barring magic preventing such, any sufficiently complex system will generate a soul of some sort.

[Harmony Ltd.]

I'll give some thoughts to that once I'll have finished reading Ch.65.

For now I just had to share a thought:

If you have a sufficiently large cybernetic body (Sasha, or cyber-alicorn Blackjack), why not put a micro-generator or three in that shit?

You'd no longer need to ingest gems to get power. On the other hand, you'd start running the risk of over-heating.

"These are not wings. These are radiators."

[Harmony Ltd.]

So, Discord's ultimate fate is now accounted for.

*strike a line from his notepad*

Godspeed, you magnificent bastard.

[Harmony Ltd.]

Re souls, from Ch.65:

PH spoiler:

So...

'growning her own soul'. This can be interpreted a number of ways, but...

- Echo's soul is only a tiny little mote of light.
- Blackjack's soul is a mote of light bigger and brighter than the average mote
- Luna's mote is a giant, very bright mote.


This would seem to point toward souls being qualitatively and quantitatively different between each others.

[Harmony Ltd.]

Though this doesn't contradict the side of your idea about how things gain a soul as they gain in complexity.

Why... Maybe it's possible Elusive could even be growing one himself?

Now that's a scary thought. Especially when you think about the shape it would take...


Though this gives me new ideas on how you could "break" him, if that was the case...

[O. Hinds]

For now I just had to share a thought:

If you have a sufficiently large cybernetic body (Sasha, or cyber-alicorn Blackjack), why not put a micro-generator or three in that shit?

You'd no longer need to ingest gems to get power. On the other hand, you'd start running the risk of over-heating.

"These are not wings. These are radiators."

Aye.  As a compromise, you could use a generator with an output low enough that overheating wouldn't be a problem, but then you'd have to rest periodically to let the generator recharge your batteries for the next activity period.

So...

'growning her own soul'. This can be interpreted a number of ways, but...

- Echo's soul is only a tiny little mote of light.
- Blackjack's soul is a mote of light bigger and brighter than the average mote
- Luna's mote is a giant, very bright mote.


This would seem to point toward souls being qualitatively and quantitatively different between each others.

Aye.  That's one of the areas where Somber's soul mechanics and mine agree, though it depends on just how the word "qualitatively" is being used (in my system, you have to do more than just changing sliders to make one soul look like another, but they're made of the same stuff).

Though this doesn't contradict the side of your idea about how things gain a soul as they gain in complexity.

Why... Maybe it's possible Elusive could even be growing one himself?

Now that's a scary thought. Especially when you think about the shape it would take...

Any sufficiently complex machine intelligence would.  If computers made out of meat and programmed by evolution and culture get souls, why shouldn't computers made out of metal and gems and programmed by coders?

Though this is one of the points on which Somber and I disagree; in his system, as was related in the conversation and understood by me, a machine getting a soul would require something similar to show!Twilights alicornification, and AIs have to make do with simulated souls if anything (It's related to natural biological intelligences receiving their souls a very small amount of time after conception, which AIs don't have... Though, IIRC, Somber did also say that an AI really slaved over and loved by its creator might get one. I'm afraid that I don't understand Somber's system especially well and am likely not doing a good job explaining it.).  Still, PH was already only mostly canon to this universe, so a few bits of mental broad-stroking to the Cognitum stuff ought not to be problematic.

Though this gives me new ideas on how you could "break" him, if that was the case...

Oh?

[Harmony Ltd.]

Well, maybe something akin to Rarity's figurines, minus the 'soul-jar' part, on a megaspell scale, applied in such a way that each nodes get its own soul-shard?

I dunno. You'd have to dwelve deeper on soul mechanics and how it applies to distributed artificial intelligences before suggesting a theory that is more than idle speculations.

[Sponsored content]