In the discussions I've had in the discord, it came to attention that there are plans to emplace Athena into other theaters, such as ground forces, the industry, and perhaps other places, I propose a robust system to make sure that it would function in a manageable and efficient way.
The current system we have the entire network be one system. Every individual piece of equipment assists by "donating" its processing power to the center, and works together with every other piece of equipment/node/etc to be smarter. The immediate group theoretically cannot communicate eight others, not in immediate range for more power intensive problems, but they could still operate independently, just not as powerfully. They also pool their supposed knowledge together, so the more they become, the more "skilled" they become.
This can be prone to several problems, although currently, they are all theoretical, I'll list them below, and I will add to the list as it comes by
- Potential conflict in how the system functions: if two networks separated for a period of time come together; The way they run could have changed drastically, specializing towards the systems that they run and nothing else, I don't list HOW they could change because I assume there are multiple ways of running a Futuristic AI network that can be better at certain tasks than others at the minimum of Structure ( Look up types of LAN Networks; Ring, Star Mesh). This is assuming that not only is it more efficient to run certain networks in different configurations, the systems themselves would be fundamentally specialized towards different tasks (the tasks they were designed to do as part of a collective). (One may act as a weapon platform, another may relay orders, while another processes Strategic level intelligence) which would require different networks based upon that system.
- System Drift: This goes back to the conflict in how a system functions. If a Network of systems is hypothetically separated, they may over time evolve into something fundamentally different than to what we have, which may cause incompatibility to our overarching network. We cannot assume that they will be compatible by default unless we force them together. This is assuming that the system is not currently designed to assimilate relatively exotic systems without any chance of malfunction, permanent knowledge loss or inefficiencies..
- Chain of Command: From what I gather, The system is not designed for "individuals" to exist. Not individuals in the sense of thinking beings, but systems "individualized" for a role. This means that there could be a potential conflict where information could be sent multiple times or relayed to the wrong systems. This is assuming that the System does not correct itself and is smarter then I am giving it credit, but I know for a fact that it is not a General AI, Yet. Please don't say "Its an AI" and throw the argument away, since I am assuming that just like a human mind, It is not able to multitask effectively enough to micromanage every single "node" to be an efficient system, because it has better things to do; It would be far better to just build an efficient protocol in the first place.
- Vulnerability: If the system experiences significant loss to the unit operation or even hostile attack in the form of info warfare, the entire network is at risk, with the potential of the entire network either failing or being subverted. It is also possible that with a distributed network losing a large portion of its units, critical knowledge may be lost on a large scale if certain units are lost. If it gets hacked by the foreign power(s), the entire network could be exposed to their ilk, even if it is an AI system.
- One-Path System:Simply put, the system does not appear to experiment with alternate or otherwise purposely commit to changed "personality" or operative procedure as well as it could do. This could be accomplished by allowing some systems or a subgroup of the entire network to temporarily experiment with a different layout than the rest of the system. As far as I can tell, this is impossible to commit under our current system since it is not designed to commit to such a task.
I propose that we commit to several changes to the Athena Network, namely to address the issues listed previously.
Revision/Design: Athena SubStructuringThis changes the Athena system by giving it three separate layers of structure, These are now fundamental to the system. The Composition of these structures, however, can be changed as is how connected they are to another Group/SubGroup/ Individual. This also consists of the proliferation the Athena Network onto every available system we have IF it is a design.
0.
The Athena Network: The entire network is connected together, If deemed needed by the network, the groups/etc can be so interconnected that there might as well not be any significant distinction to them or only through the minimum of being able to request information from one another, but the ENTIRE broad network is a Meshed system. Every single Group is connected together, at least slightly.
1.
Group Layer: The Athena Network is divided into distinct, whole groups of systems that run in the same Network Theatre. This means that groups of systems (Void Military, Void Logistics, Void Industry, Ground Military, Etc) will operate in the same overarching network, and will share group level information that everyone in the entire network needs, or work together for strategic level thinking and planning. This partially solves the Vulnerability problem, as it means another later of networking and security that a potential intruder has to access before they can acquire production secrets that allow us to produce our equipment. This also increases the efficiency of the system, as only relevant data will be stored in these groups. These groups will self-determine how each subgroup and slash or individual connects to one another, along with their priority in their messages and will also distribute "imprints" to other systems that can receive it. Imprints will be discussed in the Individual layer. Right now Athena, as it is in its current form, would fit in as a single group, With subgroups of fleet.
2.
Sub Group: These subgroups are where information and processing power is shared on a more local level, or where critical information is stored behind additional security or other specialized roles that emplaced onto more than one "node". This also allows experimental organizational patterns, network structure, security/ "Software" tests can be fielded without risking the entire network, as its specifically designed to allow the system to self-innovate along with considerations in terms of long-term stability and structure. This allows anything from groups of Uranium Centrifuges to harmonize together, Laser Point defense systems on a starship to target the same place, and everything in between. Subgroups require a MINIMUM of two units, otherwise, they are simply under the main group as an "autonomous" unit. One Unit can be under multiple groups (Scout, Infantry, Experimental, etc) If required, adding or removing groups is designed to be very fluid and easily done, limited by the information transfer of the necessary knowledge that the subgroup requires. There can be Subgroups within Subgroups
3.
The Individual: Each individual is programmed with a very base level set of "preferences" in their day to day operations, with a slight range of preferences towards certain tasks and "Mental" processing. In the scale of the entire network, this acts as the overall "Personality" of the chosen Group/SubGroup. If a query to an entire subgroup/group where Personality based ont he Imprint gets to show is given, Each available Unit answers, and the answers of all machines are composited, and then outputted. This is how divergent thinking is both reigned in and able to be given to the systems, as it gives a chance for potential shifts in thoughts as the base units can learn from each other as well as other information. Imprints are basically the set of values that the unit in question has, and is often overridden periodically with superior imprints founded by hard won experience. Imprints are inherently different for different systems, A imprint for a theoretical spaceship will not fit on a tank or an industrial centrifuge. Worryingly, the Network can and will "weld" these values onto a different system for testing purposes. In one particular instance, this has resulted in a Laboratory Kitchen unit with the mental imprint of a starship astronavigation computer, which promptly self-terminated. Systems with more overall power in their processing or otherwise more access to equipment will have a stronger pull in what to do then a lesser machine in (sub)groups/networks.
How is this better than our current system?
It provides Security, Durability, and function while being more efficient with an increased ability of self-learning and innovation, and it is also distinctly similar to the system of our government if our system of government was less like a system of government and more a tactical smart network.
The Security is given two-fold. The most obvious fact is that there are multiple barriers between a theoretical subversion of one of our individuals and ALL our individual systems. There is a significantly increased amount of data checkpoints that could be placed than as one permeable layer. It's not likely that our AI could get subverted as a whole, but individual units could fail, and certainly, more could follow if they attempt it. It also provides "structural" security, in the event of large-scale casualties, as large-scale number loss poses less risk in lobotomizing the network (Violent network failure due to nodes failing to communicate resulting in a paralyzed system.) Because it is likely that only Subgroups would die in whole scale numbers, (A squad of 7 getting shelled, a daring fleet squadron getting destroyed, An industrial accident taking out all the alcohol fermenters, etc.) While the larger "Group" network will still be relatively whole.
The Athena system will also be more durable, largely because of the previously mentioned number loss issue. Where it really shines is that it becomes more of a smarter, tactical/strategic network then what it originally was. Without the gigantic task of having to mesh with any non-similar system at any time, the system will be less prone to issues relating to connection failures or exotic systems, because they will be connected at a "Higher" level, rather than from system to system. (Kind of like how multicellular life work.) It is also more readily able to handle exotically structured networks or other systems, as they can be neatly categorized and kept in their state in a subgroup or even their own group if required.
To note, Although this is sectionalizing and "dividing" the Athena network, The network will still have the exact same resources and intelligence that it would normally have, it just organized more sensibly. All the computers are still connected in some way shape or form, and the AI can change how it is connected together specifically. Every system in the network can still pool together to solve a gargantuan problem if they can, but inefficient systems and irrelevant data are either pruned and sent to the proper Group or sent to the Archives back on the homeworld. Athena can still act as a single system, but its separate parts can act as well.