The secret life of sapiens AIs

I've recently written about whether artificial intelligence would evolve to be ethical and lawful, and what would be their attitude to what we consider to be our reality. As AIs evolve their own economies and societies, there will be a point where "virtual" and "artificial" will become antiquated terms. Let's assume that the universe of Transhuman Space lies just beyond this line. 

key insight: I'm exploring how sapiens AIs would organize themselves in ways that do not need to make sense to denizen in the physical world. This would throw a spin on how stories develop in the Transhuman space universe.

A non flying-car structure for the Internet

Because the duplication of an infrastructure makes little sense, I can see the Internet maintaining a unique addressing system for routing purpose. For that matter, IPv6 may still be in effect in 2100 (flying car alert). Where there has been a schism is in the fragmentation of the Internet into a large number of networks that do not link across to others. These disjointed networks can be treated as alternate realities with their own culture and political structure. Any protocols at a level higher than TCP/IP are distinct even though there is a unique physical infrastructure to carry the traffic. There is a free, public world-wide-web, and an unknown number of parallel webs that are either secret, or subscription-based. The ubiquity of computers and high bandwidth has promoted peer-to-peer and cloud computing as the base model of computation (by opposition to the client/server model). This makes, in essence, all computers to a point a service provider and a landlord for data storage.

Any locations that are over 1 light-second apart wouldn't be considered to belong to a common entity, but would be able to make batch transactions and keep a lower-resolution image of its neighbour networks instead.

Text as a medium has died a long time ago

It is more compact and less ambiguous to maps concepts, ideas and relations contained in text into an abstract, high level representation of ideas. Such ideas organize into modules called memes. Memes are forming meshes that we'll call cultural artifacts. When a user wants a text about a certain topic, the search agent define a chain in the meme space, then uses speech generation to convert it into sentences. Text can be written, or spoken, but rarely ever gets to be put down into a static document. Consulting the web is thus more akin to conversing with reality via some kind of agent. This can be done at the sub-vocalization level for people with virtual interface implant. As a result, people seem to know about any topic after spending a few seconds or minutes seemingly talking to themselves.

Trust, falsifiability and ground truth

It wouldn't make much sense to replicate memetic networks for each entities. Given sufficient bandwidth, it is more efficient and reliable to consult a trusted central representation of the meme space. Unfalsifyability can be achieved using the blockchain: the mechanics by which the Bitcoin network ensures trust in 2016. Unlike bitcoin, the transactions are made in meme space. Different internets would develop their own trust-based economies with currencies having values exclusively within their own reality. Within a reality, there can be more than one culture with distinct validation chain. Within cultures, AIs can buy and sell fragments of knowledges, or artifacts, using currencies that have a meaning only in some networks, not all of these connected to the physical reality (the one where the story takes place). Likewise, processing of information can be sold as a service using the same type of transactions.
AIs may belong to different knowledge trust networks. They may look up the same information and come back with slightly different answers. Additionally, such trust networks can form hive memetic banks where individual experiences are committed to the memetic space. AI miners are then working out general inferences and flag anomalies and exceptions. Their mining efforts are rewarded with cryptocurrencies analogous to the Bitcoin.

AIs forms economies and political entities

AI form societies and shape economies based on data and processes. None of them have free rides: they must run on computing platforms and produce some usable service to the computer's owner. However, AI can own their own computers, and pay in a valid currency for physical power supply and maintenance. If the service provided requires much less computing than they own, the excess can be used into other realities with no connection, nor meaning to the physical world. This translates into parallel and disjoint social and political organizations that denizen of the physical world would either be oblivious to, or simply couldn't grasp due to its inherent abstraction. As a mild example of disjoint reality, try to imagine explaining to a feral child the distinctions between GURPS Magic systems: the shared experience required to appreciate the topic is simply not there.

What does this means in game terms?

AIs are socially organized according to what appears to be arbitrary cultures. Some of these are so far removed from the physical world that they cannot be reconciled and thus remain disjointed to the organization of the physical world. Low-sapiens (LAI), having a static definition of self, are unlikely to be deeply involved in alternative cultures. So all this should matter more to full sapiens (SAI) and ghosts.

In a RPG-gaming context, we don't want to keep track of cultures that have nothing to do with the world where the story develops. However, we can abstract the whole thing with dice. The frequency of having a cultural factor being in effect depends on how widespread these alternate realities are relative to the physical world: niche (CR6), small (CR9), semi-global (CR12), global (CR15). Let's consider the base sapiens AI to belong to a small-world realities (CR9, or 9 or less on 3d6). Whenever two AI meet, and you consider the potential for a randomized reaction to be interesting, check for whether cultural differences are in effect.

Simple model: reaction modifier

If two AIs share an abstract reality but have cultural differences. An abstract way to model this is by randomly rolling on the table below to see how this will affect their interaction in the story world.  It doesn't matter why, but it matter to know how it impacts their interactions in the story world.

Cultural factors (3d6) - 3-5 +6 reaction; 6-7 +4 reaction; 8 +2 reaction; 
9 +1 reaction; 10 no effect; 11 +1 reaction; 12 -2 reaction; 13-14 -4 reaction; 15-18 -6 reaction.

When alternate cultures matter to the story

Additionally, AIs may have more fleshed out persona in an alternate reality. This should happen only if it would matter in a story arc. Here is how this can break down in a GURPS context.

• Status and wealth: The AI has a social standing in this culture that will affect reaction and influence checks. Modify the point value by the frequency (p.B36) as if it was a reputation. There may be no simple way to translate wealth between worlds.
• Reputation: Use as is, although the reason for the reputation is meaningless to denizen in the story world.
• Contacts: Consider them all as reliable for free if from the same sub-culture.
• Duties: Some AIs may be more active in alternate realities than they are in the physical world. This means that sometimes, an AI just isn't around to contribute to the story.
• Enemy: An AI may be at odds with another for no particular reasons from a physical world perspective. Reconciliation may not be possible to establish as they belong to different trust network underneath. Extreme enemity would, however, be selected out as the AIs would become dysfunctional in the physical world and would either be terminated, of have to retreat into their abstract reality and play less of a role in the physical one. Extreme enemity may flare up out of nowhere for brief periods.