# Cost-Benefit Analysis of a Future Government faced with Affordable, Readily Available Super-computers to the Masses [closed]

Premise

I'm imagining a world where super-computers are available and moderately affordable to the masses.

In 2017 alone, there have been several cyber incidents that have cost state and private sectors around the world billions of dollars. It's clear that cyber-threat is one potential cost. Jamie Bartlett, a renowned digital technology expert, recently commented on the role of artificial intelligence in a forum held at the Oxford Union. Bartlett made a case that AI would likely help cyber "attackers" more than it would the "defenders." Basically, it would be like "trying to catch raindrops." If the world has 1 billion AI capable systems, that's 1 billion more points of attack. So, I can understand his concern, but only up until a point.

The second part of my premise is to use a close historical analog to analyze potential benefits. In the early 80's the idea of a "personal computer" was virtually an oxymoron -- what on earth would you use it for? At that time, only the scientific elite had the incentives, technical aptitude and budget for a computer (don't forget at that time, they were enormous, room-sized machines). Of course, as technology fell into place, it was no longer cost-prohibitive and the masses became tech-savvy relatively quickly. This essentially created a large demand for PC's within a few decades.

On the one hand, it's true that the risk to infrastructure and data due to disruptive events in cyber-security is an adverse effect of the decentralizing (making them more accessible/affordable) computers. Yet on the other hand, if we never decentralized computers, we wouldn't be in the advanced digital age that we are in now, and the economic loss implied would be incalculable. So who's to say that the masses wouldn't find a revolutionary way to transform the world of business by embracing affordable super-computers much the way they did in the early 90's and 2000's with PC's?

Assumptions

• Cost: Supercomputers are not overly cost-prohibitive (think the middle class can afford them)
• Computing Power: These future super-computers I am imagining are the size of a large desktop PC, but have the computing power of a small data center
• Type of Computing Power/Architecture: high parallel processing as well as single task processing

• the outcome of the cost/benefit analysis alone will determine policy (no exogenous factors)

• citaris paribus (assume people are equally as greedy, innovative, ect in the future as they are now)

*You may or may not agree with these assumptions. These assumptions are merely here to keep the question within a reasonable scope.

Question(s):

Even if super-computers were affordable and available to the masses, there is still the issue of policy. How would a rational state actor perform a cost/benefit analysis of the legality of owning a super-computer? Which tipping points would they look for? Would the tech giants resist? Would the state actor have any vested interests one way or the other?

These questions are all interrelated, you may choose to focus on any one specifically or address the larger picture as a whole.

• To a large extent, "super" computers are relative. If computers with that much processing power became common-place, it would probably be because that level of power was needed for most applications, and the new super-computers would make the super-computers of today look like... well, old IBMs. – Xenocacia Nov 21 '17 at 7:25
• Just for completeness' sake, are you talking about conventional computing? (That is, computers like we're used to them today, just faster; not quantum computers.) – a CVn Nov 21 '17 at 10:55
• Last decade's supercomputer is this decade's desktop and next decade's mobile phone. – Separatrix Nov 21 '17 at 11:30
• The question is ill-defined: you do not give a legal definition for a supercomputer; in order to pass a law prohibiting private ownership, the legislators must first define them. In the early 1980s computers were not necessarily very expensive room-sized machines; you may want to look up the Apple II, introduced in 1977. I am puzzled by the sudden interest of politicians in artificial intelligence. We currently do not have the slightest idea of how to begin thinking about making a general-purpose artificial intelligence. – AlexP Nov 21 '17 at 14:22
• Got them now. For a few $K, you can get a dual CPU motherboard, 2 16-core I9 CPUs running at 4+ GHz, a pair of NVidia GPU boards, a largish SSD... – jamesqf Mar 29 '18 at 5:30 ## 4 Answers We're already there. There are people building supercomputers using the Raspberry Pi. Every generation of computers (which is less than a generation of people) changes every level of computing capability from top (data center) to bottom (personal). The first few Cray computers - fastest in the world when produced in the 1980s - are easily beaten by a typical smartphone. There are some challenges, but consider Bitcoin mining and distributed projects such as Folding@Home and you can see supercomputer-level computation happening "everywhere". Restrictions - policy & outright legal limitations - only go so far once the machines are out there. You can restrict the top-level standalone supercomputers easily enough. It is a lot harder to block use of "ordinary" 3 Ghz. multi-core processors and even harder to block people from gathering together their$5 Raspberry Pi Zeros that they can smuggle easily anywhere. Today's mainframe/data center/supercomputer is tomorrow's desktop is the next day's pocket computer.

The key question is encryption. That, or really decryption, is the holy grail of supercomputing. Brute force can work with today's inexpensive computers to crack, at least in theory, encryption from relatively few years ago. If it really takes quantum computing to truly crack current high-level encryption then an affordable quantum computer becomes the device for governments to attempt to restrict as if it were a weapon. But short of quantum computing, this is just a never-ending race of technology, human factors (e.g., phishing attacks), etc. as has gone on throughout the history of cybersecurity and really in many ways for much longer than that.

Looking at the present, we see the governments of the world seeking to limit or control the access to information to the masses of people. China is the most notorious example, with the "Great Firewall of China", banning VPN and forcing people to use the government approved versions of Google, Facebook etc.

However, Western governments are not above meddling with information control either, for example EU rules involving so called "hate speech" on social media platforms (with very fluid and elastic definitions) can be readily abused (and indeed there are already accusations of just that). Even non government actors can use their positions as quasi monopolies as "gatekeepers", Twitter's "Trust and Safety council" bans, shadow bans or limits user accounts according to arcane rules which are often accused of limiting the accounts of conservative or right wing users. Google has similar policies which are also accused of "de platforming" people with conservative political views. Facebook is accused of manipulating newsfeeds to prevent certain topics from "trending".

The rapid availability of supercomputers restores the balance between individuals and governments or quasi monopoly platforms with large amounts of resources. Gatekeepers generally have asymmetric advantages over individuals in terms of resources. In the 19th and 20th century, mass media like newspapers, radio, movies and television required elaborate and expensive equipment to produce, and expensive distribution chains (or limited and controlled ones) to distribute their messages. Only people with access to similar printing presses, studios or transmitters could compete.

The process began breaking down with devices like photocopiers allowing the creation and distribution of quality print materials at low costs, challenging newspapers from the bottom up. The introduction of the Internet to the public in the 1990's did the same for all media.

Large scale access to supercomputers (like today's Nvidia machines) will tilt the balance back to individuals again. The results will be difficult to fully predict, governments and large corporations can buy supercomputers in bulk, creating gigantic "nodes" of supercomputing (supercomputer clusters), while individuals could create ad hock networks of supercomputers in shifting and ever evolving clusters and meshes, which could provide local information superiority, and would be more resilient against network attacks or even damage to infrastructure. At some points in time, the large players could have advantages (i.e. China) until the community of users finds work arounds to breach or bypass organizations in gatekeeping positions.

The damage done by cyber attack is the loss of data and the loss of digital resources and the loss of real-world utility that flows from those digital resources.

From the perspective of the individual citizen, I'm not sure there is a point at which it becomes a rational choice to deny use of digital resources in order to avoid the loss of those same digital resources. It would be like burning down your house to avoid getting burgled.

From the perspective of the state, it may be that the gain in power that the state receives vicariously by its citizens having computing resources becomes more than offset by the loss of power from the state to its citizens.

(a minor quibble: In the early 80s the idea of a "personal computer" was very-much not an oxymoron. The Commodore PET, Apple II, and TRS-80 were all introduced in 1977.)

I'm not sure a rational state actor would be capable of performing an accurate cost-benefit analysis, any more so than world governments were able to predict how the personal computer, then the Internet changed the world.

But assuming they did, what should they look for? - security: information security has three axes - availability, privacy, and accuracy; there are data protection techniques other than encryption that span this spectrum. A computer does not have to cough up data just because someone has provided a password. It can refuse to answer questions unless a host of authentication challenges (factors) are met. Data at rest or in transit can be protected using algorithms that rely on the algorithm's secrecy, limiting who may participate in the exchange.

Current encryption technology only does the high powered encryption on the key part of the message. The bulk of the data is stored using the simple key that is better protected. The reason we do this is because it is computationally expensive to do otherwise. However, in a world where more computational power is available, then every bit could be protected. Also, of course, we can add bits to our keys making brute forcing take even more time.

A good security analyst could do a more thorough job here, but there are plenty options available. I would not assume that computer security is hopeless.

Contrast against possible growth. Even if unprotected, additional computing power opens up huge new horizons in chemistry (predictive modeling of reactions and the ability search for reactions that might yield materials with desired properties), engineering (more realistic environments to test vehicles and structures in simulation without the cost of prototyping), art (computer generated music (sadly), performers, maybe even whole products), medicine (better ability to model health and drug effects), genetics and genetic engineering using things like CRISPR, civics (AI-aided social networks), city planning, and finance, just to name the first few I could think of.

I think a rational state would recognize that it is incapable of evaluating either the cost or benefit effectively and just go with their guiding principle - be it hands-off or micromanaging.

Regarding tech companies: they are currently positioning themselves like ABC, CBS, and NBC did back in the radio age, as gatekeepers of the technology. They will seek the government enact rules preserving their dominant role. Depending on their success in this regard, the center of innovation will move to whatever geographic region provides the most minds the lowest barrier to entry.