Topic: LogLore: Technology

[big]LogLore: Generating Seed lore through discussion[/big]

The following is a formatted version of the discussion between the GM and Mike:

Question: What is an AINN?
Answer: So, the idea of an AINN is that it's the smallest autonomous computing unit in TAU. They can perform various functions on their own, like the functions of a small factory, workbench, Chimbot or tool, but their true potential comes when networked. When networked they form TAU, a swarm intelligence. All the AINNs contribute to the intelligence, though some are specialized to AI heavy processing tasks. Most AINNs contain a computing unit, some memory (temporary and persistent), and interfaces to various other circuitry in whatever machine they're embedded in.

Q: Are they like nodes in the supercomputers of present-day Earth?
A: Yes, but in present computers the hierarchy is rather flat, although there are control systems that give the nodes things to do. They are scalable such that nodes can be added and removed, and AFAIK a single node can also function on its own (hardware-wise, but I'm not sure if they usually have software to use anything useful without the control system) They use multi-gigabit backbones so wireless technologies don't quite cut it.
    No, obviously TAU's software is much more advanced. The AINNs are hierarchically organized so that some nodes give orders to others and they can function in a way if the nodes are separated. AINN swarms (defined as groups connected to each other at any given time) are also hierarchically organized in the sense that they can determine a "senior" process that can give orders to others. Originally, this role was held by 'TAU Master Processes', but after the disconnect a new senior process was determined. It resides mostly in the Middle Cluster AI Core in Labspace (between the two levels).

Q: What is a 'process' in this context, a single program running on a single node or like a higher level concept, more like 'thought process'?
A: I guess it's a higher level concept. At least it can run on several nodes at once. In the more fractioned areas, there's often practically only one relevant process present anyway, so that itself is the senior process. For example, a disconnected Chimbot walking around semi-randomly, looking for points of connection.

Q: How do the nodes communicate?
A: In contrast with present day technology, where wireless or wired connection is the physical layer and communications protocols like HTTP and SSH and others are above that, the basic connection is done by hardware. Wirelessness is such a norm at this technology level that it's embedded directly into the basic hardware. There may be wired or wireless network interfaces, but the processes don't need to care which is in use. Cost probably isn't an issue when the mission is to colonize another planet, and it wasn't much of an issue in the Technocracy either.

Q: Can you give some examples of the kinds of processes there are?
A: Well, for example, a process might be a factory's process for producing the items it's producing. Or a scientific simulation. Those processes can also have subprocesses that look more like present-day processes. Or maybe a life-support system's process for determining the current state of life-supportiveness and reacting accordingly.

Q: I suppose there must be some sort of operating system that provides communications and other infrastructure to the processes?
A: The operating system's a good question. Naturally, there's one.
*joke* Yes, of course all of this uses Windows 5000 as the operating system. Not really, that's because in 2113, too much Linux usage collapsed the world economy, and all such things became history. *joke*

Q: Is it possible to interact with the OS through some sort of interface (shell)? Like, examine the files and such?  Or is all of this hidden behind abstraction layers?
A: It is possible to interact with it, and often done too.

Q: Is there any sort of local interface for the nodes?  Like, if you open a maintenance panel, what do you find?
A: The interface with individual nodes is remotely by necklink (default and intended) or RCI (hackerish, more versatile than necklink). There's no physical interface. Some machines have physical interfaces too, though. But they're not usually particularly close to the AINN.

Q: How does the interface present itself?  Some sort of graphics projected to the user's mind?
A: Yes, the user interface. As for most functions, the operating system interface is highly customizable (the interface code is in the necklink). There are visual, aural and conceptual interfaces that all are displayed directly in the user's mind, though an RCI also has a more traditional screen option.
The conceptual one is something akin to pure thoughts. Though it's very crude so it's not called telepathy. Crude in the sense as in actually transmitting a person's thoughts and feelings. Quite exact enough for technical matters. For additional details see: … 3623#p3623
The visual and aural versions also don't obscure what you actually see but are instead projected into the "mind's eye" (or ear).

Q: One question about electronics in general - what do they look like?  Are there still circuit boards?  Are things modular like today's computers?
A: (somewhat hectic, go figure...) Atomic circuitry too small for the eye to see. I'm guessing modularity is the theme on TAU, though not quite as much as in original Seed.
Could they look like lego bricks? Which you could combine in various ways in 3D? Would that make any sense?
I'm not really sure how to make the 'various ways' fly with electronics, where things such as signal path lengths and shielding are essential.
Well, 'lego bricks', 'various ways' and '3D' make me think that they could be assembled in different shapes. And I can't figure out how there could be enough connectors for that. A multi-tiered hierarchy does work, as long as the connection points are clearly defined. And the components in one unit (equivalent of a board) arranged in 3D instead of on a plane?
I'd figure that'd save some space. And wiring length. That's where things like signal routing become difficult. I suppose it is possible if each component has passthrough connectors for chaining other components on each side.  Perhaps future technology has a solution for keeping various signal anomalies such as noise and reflections down. With today's technology, connectors tend to be a weak point of the signal path, reducing quality. They don't have as good shielding and the wires can't be arranged in a way that provides maximum shielding from interference.
The component is connected to its parent through one connector, and a passthrough connector's pins are connected directly to this connector's pins. GM: Shall we call 3D electronic components feasible in the future? Without lego level modularity but instead with defined interfaces. Yeah, they are feasible, as long as the connectivity doesn't get too complex."

Q: What causes AINNs to disconnect?
A: It's caused by high levels of background noice. Wireless connections are rather suspectibe to interference, so it doesn't have to be lethal levels before that's feasible, considering obvious boosting actions. (Not that it isn't lethal levels at some parts.)
You can't just put a huge boost device on your laptop to get reception anywhere you want, becouse the signal strength degrades by the square of the distance, and all kinds of obstacles cause even more attenuation. If we ignore concerns about what sort of power levels are actually feasible, then yes, it is possible to increase wlan operating range arbitarily.

Q: One thing that bugs me is that the RCI is capable of probing any electronic device, even if it isn't equipped with communications hardware.  How exactly does that work?
A: By reading the radiation emitted by it and analyzing it. And the other direction by creating fields that stimulate wanted kinds of currents.I mean, if it can determine the time and location of each signal, it should be able to get a picture of what's happening in the circuitry. The RCI is high tech and uses quantum computing. The signal sniffing and induction is still erratic, unreliable and based on some guesswork.

(Some more details on the feasibility:
The translating-singnals-into-logic part requires rather huge amounts of computing power too.There are 500 million transistors in a modern CPU, which change state 3000 million times per second.  Assuming you can detect just the transistors, it requires analyzing on the order of 1e+15 signals per second.  So the analyzing processor should be still a couple orders of magnitude faster, or be massively parallel. And if you have to consider all possible interrelations of those signals...  It gets astronomical. Couple of orders of magnitude as in 100 times faster. Quantum computing just might be the technology needed to make it fly. It's massively parallel at least.)

Edited and formatted by [death]dustman[/death]