Questions: Persistent Memetic Weapons and Machine Learning

1.  Referring specifically to The Laws and Customs of War:  What exactly is the difference between persistent and non-persistent memetic / infoweapons?  It’s obvious that the big distinction is inherent in the name, but, to be more specific, how to the people in charge of using such weapons ensure that they are properly “infodegradable”?

Very carefully.

Or, slightly more seriously, this is one of those occasions on which I invoke the “I just write about it, I don’t actually have a complete science of memetics stored away” clause. But I can safely say that there are lots of very clever people engaged in threading the needle between “Oops, our economic sabotage meme-weapon got a little bit out of hand and caused the Great Depression” (acceptable collateral damage) and “Oops, our economic sabotage meme-weapon got entirely out of hand and now half a dozen systems have to put up with bloody Marxists for the next half-millennium” (very much not, and the hearings will go on forever).

(Infoweapons, by contrast, are analogous to computer viruses, etc., and as such it’s just a matter of making sure you got your termination conditions and fail-safes set up right.)

2.  Regarding Powers as Programs and Skilled But Naive:  On the one hand, part of me thinks that, if you’re able to trade the raw skill itself by mnemonesis, the same should be able to apply to the experience as well, since that in itself could conceivably be boiled down to the knowledge of “what works and what doesn’t” and the memories that knowledge is associated with, and that, given the setting’s information technology abilities, these experiences wouldn’t be that much harder to swap than the raw skill knowledge itself.

(On the other hand, while typing that out, I came to realize that the idea that “to play as a virtuoso, you still need to practice like one” might still apply in practice even with that caveat I just mentioned:  In an almost evolutionary sense, the skills of yesterday’s virtuoso become the baseline for today’s practice, so that to be acknowledged as a virtuoso now you have to push out your skills even further than before.  Is that basically how it ends up working out in practice?)

The Doctor Who must be seeping into my brain, because the first thing I want to say here is:

“People assume that the mind is analogous to a computer with an attached database, but actually – it’s more like a big ball of wibbly-wobbly… thinky-winky… stuff.”

…in any case.

The first problem here is that experiences are really problematic to swap. That’s because a very large chunk of the mind (the “psyche”, or “incrementing memory string” in the local jargon) is your experiences and the way they shape your mind.

Remember, importing the incrementing memory string diffs is what you do to merge forks of yourself back together again. Importing a whole bunch of someone else’s experience-memories will change your identity – it’s hard enough to do this with your own without running into nasty cross-link problems – which at best will be enough to cross the legal threshold and turn into a fancy way to commit suicide and become someone new, and at worst is merely the fast track to committing suicide and become institutionalizably schizophrenic all in one move.

(There is such a thing as exomemory technology, but while that lets you experience someone else’s memory from their point of view, it doesn’t actually patch it into your mind as if it were your own. You can only learn from those what you would learn from watching the violin prodigy, not from being the violin prodigy.)

The second and bigger one, touching directly on the thinky-winkyness, is that the mind is encoded in what we can call a holistic, associative manner. Everything is interconnected with everything else, and it’s those complex interconnections that make (a) it very hard to comprehend, and (b) everything go smooth.

It’s easy – for values of easy equal to ‘requiring extremely sophisticated cognitive science’ – to scribe raw data into the brain as fact-memory. It’s rather harder, but possible, to encode skill-memories, and gets even harder when you’re talking about the need to go poking around in the cerebellum and all manner of other specialized areas to teach them what they need to know to go with the skill-memories, and that in turn becomes a dozen times more complicated when we have to get into how these interact with hormones, other glandular effects, and that any given body will not respond in the same manner as any other given body even before we start talking about neomorphic shapes.  But it’s possible.

Where it gets impossibly hard is in editing in all the millions of little subtle connections to every other part of the contents of your brain that would have been there had you learnt it in the conventional manner. And without those – and this is a poor analogy – you’re in the situation of someone who tried learning karate from a textbook. Or someone unpracticed with an English degree trying to write poetry for the first time.

(I mean, you can still turn in an expert-level performance, since you have the skills, but that’s not the same thing as having them fully integrated into your self. Like the trope write-up says, it’s about integration and synthesis, about building all those connections that let you do things without having to try to do things.)

Now. That all being said – this is a technological restriction. If you have access to all the powers and power of a Power, in the Vingean sense, and thus are or have a friendly mind which is capable of not only comprehending yours in every single aspect and fine detail, then they can re-envision you as one possible person you would have been had you known these things all along and spool that straight to output. It’s easy for a Power to do that. They write software of greater-than-human-mind complexity every day of the week and twice on Nyxis.

But the gods are very busy, and have better things to do than come running every time someone wants to have learnt kung fu.



So, Gregory Johnson was asking about education:

On a tangentially related note (to firmish SF), how does education work exactly, in your universe? It has been implied in several places that skills get basically downloaded (or can be), while it still takes to age 18 or so to be educated to Ph.D. levels. What does education really LOOK like?

Well, indeed, they can be. Data-sets, skill-sets, and so forth can be downloaded across the dataweave (mnemonesis), or even pulled out of the collective consciousness (remembrance), along even with personal memory sequences (exomemories). Indeed, dynamic mnemonesis enables you to remember any data on record as if it resided in your own memory. Advanced gnostic overlays permit the download of entire partial personalities, instincts, professional mindsets, and other similar temporary mental modifications.

The problem, of course, is that just having the knowledge won’t do a damn thing for you on its own. Grab some chump off the street and download a full knowledge of, say, proteomics into him, and all he’ll get is an uncomfortably full sensation in the brain-pan. The problems are two-fold: first, learning is associative, and without something to hook onto, said knowledge will be inaccessible. Second, it is a matter of processing power. Downloading knowledge into a mind that never learned to think doesn’t suddenly enable it to, whatever its genetic gifts.

So, education – about which I am only going to speak in general terms in case I want to use it later – is much more focused on teaching people how to think than on learning facts; facts are easy to acquire. The “primary” education initially focuses almost solely on this (logic, metaphysics, epistemics), followed by a “secondary” education that is their equivalent of a liberal arts education, which provides the very broad-based core that such modules you download later can hook into, along with the lots of practice needed to synthesize the gentle art of thinking.

(It’s actually rather broader than what we’d classify under that name – the traditional strands in it could be given as Advanced Logic & Mathematics; Business, Finance, and Economics; Domestic Arts; Engineering; Ethics & Civics; Fine Arts (both appreciation and practice); History; Literature; Martial Arts (both armed and unarmed); Natural Philosophy/Science. So, y’know, that this is what they expect any reasonably educated person to have good knowledge of might explain a few things…)

As for the hows, they don’t have schools (both for reasons of population demographics and because, well, they’d be absolutely terrible at doing this kind of education, rather than the kind of fact-and-discipline-centric kind we use). Education at this level is home-based, delivered by parents, the child’s muse, and companion AIs (and, of course, mnemonetically, for raw facts). Stylistically, it’s integrated into day to day life (since learning, they find, sinks in best when it’s fun and easy). Much of it is also practically-based; children are rather more integrated into society and work, and as such learning by doing – usually in whatever eclectic things strike their fancy – forms a great part.

By the time that this period is over, its average “graduate” has the equivalent of at least a couple of degrees worth of educational achievement, albeit widely spread and electic.

Now, as for higher education, it’s similar to ours; one attends classes at (or remotely from) a university. The distinctions would be that it’s very unstructured: organized by the class, and the degree one comes out with is just a matter of total achievement, not a specific pattern; and that there is little point to lectures in their paradigm, since information is easily conveyed mnemonetically. Instead, courses concentrate on class and lab time – discussion and practice at practical application are what synthesize and integrate the mnemonetically delivered knowledge with your core self.

Also relevant reading: Powers as Programs, Skilled but Naive, Neural Imprinting.

Trope-a-Day: Powers As Programs

Powers As Programs: With the availability of mnemonesis to download knowledge and skillsets into your brain, gnostic overlays to do the same thing for professional personalities, and the ability to download recipes for your portable nanolathe or desktop nanoforge to build you the tools to use them…

…yeah, this is pretty much the way it works.  Although those powers which require genetic changes, large tools, exotic materials, or significant ‘shell modification are somewhat more difficult to achieve.

But see also Skilled, but Naïve, when we get there.

Trope-a-Day: Skilled, but Naïve

Skilled, but Naïve: The consequence of relying too much on mnemonesis and gnostic overlays to download distilled knowledge and skillsets.  They can, and will, grant you extraordinary, even superhuman, technical skill.  Excellence is guaranteed.  You will outperform any mediocrity, even any average professional you ever meet on raw talent alone, because you have been given a magnificent instrument.

But it’s the integration of the downloaded knowledge with your core self that lets you play it like a virtuoso, and there’s no artificial substitute for that.  Just experience and practice.

Trope-a-Day: Exposition Beam

Exposition Beam: What mnemonesis (see Neural Imprinting) is for, although it’s much more commonly used for imparting knowledge in the data and skillset sense than imprinting knowledge in the history/backstory sense.  Nonetheless, it can be done, and as such, tends to be done, at least sometimes – despite the awkwardness that imprinting that sort of information this way tends to come with lots of context and emotional sidebands that you might not want, because copying historical memories this way (rather than distilling them down into a matrix of facts first) means that the recipient remembers them as if they happened to him.

(Abuses of this phenomenon are also well known, and fairly unsubtle.  See Mind Rape.)

Trope-a-Day: Neural Imprinting

Neural Imprinting: Ubiquitous, although it’s called “gnostic overlay” and “mnemonesis”.  And while it is the standard educational technique (starting prenatally, with the axiom feed), it’s only the first step.  After all, being able to download huge chunks of skillset into a chap’s brain is easy; successful integration and synthesis to make said skills your own is hard, and then practice, practice, practice!

Of course, in terms of pure fact-memory, being able to recall just about anything known by remembering it certainly helps.  Assuming you have the intelligence and intellectual skills to do anything with the data, which is the hard part there.