by Vic Grout
“Yes!” Aisha could not restrain herself; Bob continued.
“The Internet is a logically organised data network, as are all the devices attached to it really. But it’s tied to another system to keep it going: the power networks. If you add those in, you’ve got a fair bit of extra complexity too. Not only that, but a power network often provides a physical connection between the main Internet and any other networks that wouldn’t otherwise be connected to it – many wireless segments, for example.”
Jenny spoke slowly, and with an air of resignation, as if she hardly believed what she might be conceding.
“And … we know … that there’s noise on the … power networks … too!” She put her head in her hands in despair. But Bob had not yet switched sides so easily.
“That aside, I think my main objection to the idea is that it’s just daft! It’s science fiction stuff. I mean, a conscious Internet; really?” Jenny raised her head and nodded but there was no other response as all four retreated, for a few moments, into their own thoughts.
Aisha quickly scanned the faces of the other three, and took a deep breath before taking her opportunity. She had now had time both to collect herself fully and refine her theory a little.
“Think about it. I know it sounds mad but think about it! In animals, the brain and body, working together, are essentially a powered nervous system. There comes a point where it is difficult to separate the two parts of the bodily process: everywhere are signals and everywhere is power. It is likely that our own consciousness comes from just this combination of signals and energy: there is no need for anything else.” A sideways glance at Andy, who merely shrugged. “The brain, our nervous system, is made up of a biological substrate on which neurons can fire data around. OK, we do not exactly know what this data is or how it does it: we can argue about whether it is information or control that is being passed. But it probably comes down to the same thing. Somehow it manages to do it, and we cannot deny that it does: we could not be having this discussion otherwise. No-one tells us we are not conscious just because we do not know how or why. Well OK, some philosophers might,” another glance at Andy, “but generally we accept the idea. Somehow, the right foundation allows for this sufficiently complex powered network to assess its own existence, then its ability to predict, then to control. It tries, it learns, and it gets better. Any notion of higher levels of intelligence must be built on this foundation somehow. In fact, if an Italian friend of mine is correct, the brain learns to abstract: the whole thing is to do with the nervous system working to a fundamental, low-level imperative to minimise its assessment of lack of control. Everything that happens after that builds from there.” No interruption so she continued.
“Now, consider the Internet, and the power networks it is connected to across the world. What have we built? We have built a massive – really massive – powered neural network. And its very function is to communicate. Every link is a carrier designed to communicate – we made it that way. Its end points are specifically designed to transmit and receive information, or control. The nodes and the links are made to work together: there is nothing left to chance, in fact.” She looked at Jenny. “OK, it is not connected in the same way as the brain but it is actually much, much bigger – if you consider all of it, which may compensate. That may also mean that it does not do things in exactly the same manner as the brain but it might do something very similar. The point is, it has all the necessary ingredients for self-awareness that the brain has. It makes little sense to worry about how exactly that might actually come about because we do not know how it happens in our brains either.
“So, in a pure sense, it is not exactly a panpsychic model, I suppose, if I understand that properly: this does not suggest that a rock or a mountain could achieve consciousness but a sufficiently complex, custom-built, powered system might. We have provided the structure and the power and the fundamental, low-level mechanisms for communication – we have built that. There is no need for an evolutionary process such as humans have emerged from: the Internet is made-to-measure as a global nervous system.” She paused and raised a finger slightly to focus attention on what she was about to say. “But in fact, we have done much more than that: we have shown it how to work!”
The others continued to listen, speechless. Aisha continued.
“Rather than just provide the Internet with the physical means to communicate, we have shown it how. It is already awash with our control and data, properly organised and packaged. There are numerous different protocols in its different parts and there is much more basic material, which we introduce too – whether we mean to or not: aborted data, corrupted fragments … and noise. It has a complete demonstration package for how to work (and how not to). Not only does it have its materials, it has its instruction manual too!
“And this is where I think it is in this process now: at a very early stage in what we might consider to be its development. Its very complexity has reached some essential tipping point in which it has acquired a fundamental internal, but global, imperative to assess control, possibly to seek to minimise lack of control. Perhaps it is that, which we might call consciousness. I do not know that I am entirely comfortable with that word but it matters little. I suspect that it does not really know what it is doing right now but it is aware that it is doing something: doing all it knows how to do – and that may be what it has been shown to do.
“So, it starts with just noise; probably quite low-level noise because – in the brain at least – we think that randomly switching between very close ‘on’ and ‘off’ states seems to be an efficient thing to do. It has transmitters and receivers that attempt to send and answer this noise internally but, on the whole that does nothing because there is no response from anywhere: it does not notice any difference, whatever it might do. But slowly, with a continued element of randomness and trial-and-error – localised genetic mutation, if you like – on a truly massive scale, it occasionally increases the difference between the two states in response to recognising some correlation between cause and effect and it gains in sophistication. This might have something to do with its equivalent of the brain’s ‘regularity spotter’ or whatever it has.” She tailed off, uncertainly. “I am not quite sure how it might do that though,” she acknowledged thoughtfully.
Bob broke in, shaking his head, also seemingly unable to believe what he was saying.
“Well, I guess the low voltage noise doesn’t have any effect at all: it doesn’t even interfere with anything. But, eventually as the level increases, it’s going to coincide with some ‘real data’ – some of ours – and the combined signal will be high enough to register a collision and force a retransmit. I suppose, over time and with enough data, it might get to recognise that and start to favour the higher voltage noise in cleaner spikes over the low-level, continuous white noise type. Perhaps that’s the stage we’re at now. It’s learnt that decent level noise spikes sometimes do something but it’s not sure what yet? I guess the same may be true at all levels: individual logic gates, device connections, network links and major data trunks?
“But then maybe, occasionally, the noise takes on a shape that resembles a data bit? After all, it’s doing that all the time with the legitimate data we’re asking it to carry so it’s a reasonable thing for it to gradually learn. A single bit also does nothing but it takes a tiny fraction of time longer for it to be rejected by a receiver, and in a logically different way, which it will notice. And, of course, it ‘sees’ all of this because it’s happening internally as far as it’s concerned. Slowly, it produces a few bit sequences amongst the noise, then some bytes, then some frames. Still none of this actually does anything proper but it still sees a difference so it’s learning all the time and it has billions and billions, trillions, whatever, of data samples to work with.” His eyes suddenly grew wide in realisation. “And that’s why we see the occasional frame fragment with the wrong protocol on a different network! It’s just playing silly buggers, ex
perimenting: learning what doesn’t work just as much as what does. It doesn’t particularly care about our arbitrary divisions: how we use the kit determines that; but it may not be limited by that eventually?”
Was this tacit agreement from Bob? Perhaps. All three now turned to Jenny. She rolled her eyes, opened her mouth wide and stupidly, offering her palms in supplication, and, so, so reluctantly, attempted to complete the narrative.
“Then, eventually,” she said slowly, “partly by sheer volume of material, partly by chance, and partly from learning what it’s been shown and seen, it manages to produce something – a compete frame, a meaningful signal, a recognisable command – that actually does something. So something comes on, or off, or resets, or switches, or breaks?” She sighed hopelessly in resignation. “OK, it’s a theory.”
Andy smiled, and said softly: “When you have eliminated the impossible, whatever remains, however improbable, must be the truth.”
But, if anything, Jenny glared at him even harder.
*
“Right: just, supposing for a moment that there’s any sense at all in any of this,” Jenny narrowed her eyes at each of the other three in turn: almost daring them to agree, “we’re talking about a single entity; yes?”
It was the following morning and they were having breakfast. They had all stayed the night with Jill and Bob. Aisha and Jenny had used the two single beds in the spare room, Chris, Heather and Ben having left the previous evening, and Andy had slept downstairs on the sofa in the living room. All, apart from Andy, appeared the worse for a bad night with little sleep.
“Yes,” agreed Aisha. “It has to be a single thing. It is connected; it is one. It would be too much of a coincidence for there to be more than one at present.”
“At present?” Jenny’s eyes opened a little.
“Well, I suppose there is no reason why whatever has happened could not be replicated. But I would imagine this is not the situation here.”
“And,” Bob suggested, “we’re assuming there’s an element of size, complexity at work here, aren’t we? We’re saying that some essential mass has been reached that’s made this happen. That’s almost certainly the whole thing, the Internet, the bigger, powered IoE; not some coincidence across more than one section. Also …”
The others looked expectantly. He continued.
“Also, Stephen says that there’s no noise on wifi – or any other wireless systems; nor on the satellite links or anything like that. I’ve checked with him again this morning. There’s no noise on the cellular 4 and 5G phone networks either. It’s everything connected as the Internet or by the power networks (or both) but there needs to be a physical connection for a network to be included, by the look of it. Separately-powered, physically disconnected networks don’t seem to be a part of it; nor does wireless itself. Yes, it seems to be a single, connected thing.”
“It?” queried Andy. “Is that the best we can do? It?”
And, it appeared that it was. Among themselves, it would never be anything other than ‘It’.
The conversation continued.
*
“OK, let me make something absolutely clear,” announced Bob, as the morning drifted towards lunchtime. “I am not calling Stephen to tell him we think the Internet’s woken up! I’m pretty sure that would be me straight off the case!”
“Is that because you do not believe it?” Aisha asked, smiling. “Or because you do not want to appear silly; or because you do not think you can convince him?”
“A bit of all three really,” admitted Bob.
“Same here, I suppose,” said Jenny thoughtfully. “Same thing really, of course.”
“Good; so let’s think about how we can address the whole thing then,” suggested Andy.
“How?” the others asked, largely together.
“Well, what can we do to convince ourselves?” asked Andy. “We have a model, of sorts. We’re not entirely in agreement on some fundamentals,” smiling at Aisha, “and we’re a bit vague on detail,” a nod to Jenny and Bob, “but we have a rough hypothesis. How can we test it? If we can convince ourselves well enough, we should have the armoury to take the argument to someone else, surely? So … what exactly is our model and what does it tell us?”
Aisha considered the question. “I suppose, in those terms, we are saying that the Internet – taken as the complete union of all devices physically connected together globally, either as data or power – generally both: what we now seem to be calling ‘It’ – has, in addition to carrying the data we have put onto it, achieved a control imperative of its own.”
“And what does that mean in practical terms?”
“Well, it means that, at the lowest level, individual logic components, possibly pairs or small clusters, have started acting as neuron comparators; assessing observation against expectation as a primitive mechanism for minimising assessment of lack of control. These are probably working within negative feedback loops, formed logically within and around these comparator units.”
“Right, then what?”
“Then, presumably, if our brain models are correct themselves, these low-level comparators cluster together to form higher-level comparators – again with feedback logic, which are capable of higher-level control assessment and more functionality, and so on, up through a hierarchy that eventually leads to coordinated sense/response, understanding; then possibly reason, problem-solving and creativity. There will be a lot of repetition, even recursion, in this layer building. But we do not know how far along that path we are with It, yet. I suspect, not far.”
“Good, so how could we measure that?” asked Andy. Aisha’s thoughts flashed back to her question to Paulo Di Iorio.”
“I suppose we would need to determine what was unique about that structure and behaviour, in a way that we could test; in a way that we could tell apart systems that were like that from those that were not. What makes our brains, and possibly It, special?”
“So, some sort of attempt at sensor-based random firing energy-minimisation, with feedback loops, but extended repeatedly and recursively across multiple hierarchical levels and units might be a feature then?” suggested Bob, smiling at his ironic attempt to imply simplicity. “That sounds like ‘self-similarity’ and ‘scale-free networks’ to me.”
“I’m not so sure about the scale-free aspect,” said Jenny. “True, both the brain and the Internet are known to have scale-free features but that’s usually talked about in terms of structure: the hardware. We already have that in place. We built that. We’re more interested in the signals themselves. You might have a point with the general self-similarity thing though, if we’re thinking about it in terms of data traffic.”
“Self-similarity?” enquired Andy.
“Self-similarity is the tendency of a system to have repeated patterns over both the long and short term,” explained Jenny, “or over large and small scales. I think it was first noticed measuring Nile river levels or something, but it’s recognised that many systems have self-similarity. Internet traffic itself is self-similar because its data tends to be in connected streams rather than independent units like old-fashioned phone calls.”
“So, that won’t help then, will it?” suggested Andy. “If the Internet’s doing it already, how can we measure if the new stuff’s any different?”
“Because it is going to be more than self-similarity in neuron pairs,” argued Aisha. “We may also see it across hierarchical levels and geographic regions, irrespective of the underlying technology: that might be a big clue? Also, we think everything will be embedded within negative feedback control loops as well. I would imagine this would provide particular patterns within comparator units as well as between them?”
“And,” Bob suggested, “we have what we could call the ‘signal to noise ratio’. The fraction of the data that’s meaningful, rather than random or useless. That seems to be changing but I guess we could measure it easily enough if we could just compare current activity with data
retained from last week or yesterday, say?”
“Yes, that might also give an idea of how much control was being registered and recognised,” agreed Aisha.
“So, can we put all this together into something measurably unique?” asked Andy.
“Well, we measure basic self-similarity with the ‘Hurst Parameter’,” Jenny explained. “Named after the guy who first looked at the Nile levels: usually written as just ‘H’. The H parameter is measured as the average difference between data points over points that are both close together and far apart: it’s calculated to be in the range 0 to 1 and any measurement much over 0.5 can be considered significant in terms of possible self-similarity. But what we need here is to extend this to look for the other features we think might be present in the brain such as the self-similarity possibly applying over both distance and logical level. There’s a kind of ‘two-dimensional’ relative similarity here possibly?”
“Also,” prompted Aisha, “we need to somehow factor in what effect the feedback logic will have.”
“And the signal to noise ratio,” Bob reminded them.
“And, once we know what we are looking for,” Aisha suggested, “we may be able to take measurements from actual brain activity so we know what kind of values we should be getting if It is behaving in a similar way. It is possible that Professor Di Iorio may be able to help with that: it is more his line of work than mine.”
“So,” Jenny breathed deeply, “if we can somehow combine the 2D self-similarity scores with the feedback control and the signal to noise ratios, then normalise this against known brain figures, we might be able to formulate an ‘extended Hurst Parameter’, which we could then use to test for similar activity in It?” She laughed. “Not much then!”
“We’ll call it the ‘S Parameter’,” joked Andy.
“Why?” asked Jenny doubtfully.
“The ‘Smith Parameter’,” explained Andy.
