Maidsafe is a cluster super computer, can it be used for "evil" and "good"

An idea as a human interest piece for the safe magazine

Q1Can the safe network be bridged to the internet where a hacker could use the safe network resources to create a super computer using a app to create a attack?
Q2 Is the safe network isolated from the web apart from the ability to use a browser plugin on your local machine?

flip the coin.

human greatest problems could use the safe network app to solve the greatest mysteries of science like,
artificial biology of man kind simulation to cure disease by creating new gene molecules.
or the question of zero point energy.

flip the coin again.

The government could use all it resources and create a super surveillance computer to control all behaviour freedom and rights.
eg. in public place and drop a piece of rubbish on the ground AI could inform you of your crime without human intervention.

It will take a long time for governments to start using Safe even if it quickly becomes successful. Universities and other researchers have the opportunity to take advantage of the network much earlier. And in the long run Skynet, excuse me I mean friendly AI, will come to dominate in terms of computing functionality. The biggest challenge will be to prevent artificial intelligence from running amok.

“An intelligence explosion is the expected outcome of the hypothetically forthcoming technological singularity, that is, the result of man building artificial general intelligence (strong AI). Strong AI would be capable of recursive self-improvement leading to the emergence of superintelligence, the limits of which are unknown.” – https://en.wikipedia.org/wiki/Intelligence_explosion

1 Like

I don’t think SAFE has any "computing power’ aside from counting SAFEcoin. AI is not an included feature. It stores and retrieves files and keeps track of who is allowed to store files and who deserves rewards for retrieving files. That’s all. So the Question 1’s answer is no.

How? The whole point of SAFE is to make all our data anonymous and indecipherable to the government. It might make it easier for the government to secure it’s own secrets… But it evens the playing feild because anyone’s secrets can be a safer than the governments… (Remember the governments are getting hacked left and right. Surveillance is far more possible in a SAFE free world than a world with SAFE.

1 Like

But isn’t distributed computing power something that will be possible on Safe? And with exponential progress of technology that will allow computing on increasingly massive scales.

Future desire to implement this.

Note: That it is going to be anon computing. That is if I set a compute task to be done, noone else will know about it and the desire is that as far as possible even the CPU doing the task won’t know what its computing (purpose).

It is the opposite to array, distributed, network computing in that it is neither coordinated or communicative. Thus the notion that it will enable a massive AI on its own is wrong.

Only if someone/group has the coinage, will they be able to create a massive coordinated computing platform. But that is still a single (set of) account, not the combination of everybody’s compute power. Mind you with that sort of coinage they would be better to spend it on dedicated computing (rented or not) that is better suited for the task.

2 Likes

I thought about it a bit, and I believe the farmers could have a computing resource that will allow true distributed computing power!

For example, JavaScript (JS) code can be sent to the network, and a random farmer receives it and executes the code in a sandbox JavaScript runtime environment. One problem is how to determine how much CPU power is actually used. And how to ensure that the farmers actually are running the code?

To prevent a single farmer from cheating, the JS code can be sent to two farmers. This might seem like a waste of computing resources, but remember even for files there is duplication (actually at least 4 copies). So a duplication of the computation is acceptable when it can provide true distributed computing.

And how to determine that the farmers actually run the code in a correct way? Since two randomly selected farmers are running the same code, the result they produce can be compared, and if the results differ, then neither farmer receives any safecoin reward. In this way farmers are incentivized to behave correctly. EDIT: More security than this is needed though since farmers can collude to produce the same fake results. EDIT2: Or maybe not! If most farmers behave correctly then there would be insufficient chance of colluding fraud farmers to “meet” each other randomly.

And to ensure that the farmers attempt to run the code as fast as possible, the farmer who produces the result fastest will receive a much larger reward than the farmer that runs the code slower.

Another problem to solve is: for how long should the code be allowed to run? And one solution is to have a fixed maximum limit in the system of say 1 minute. If the farmers can’t complete the execution of the code in time, the user who sent the code is charged a certain amount of safecoins.

In this way a single client can send thousands of JS snippets out to farmers who run the code and return the results to the client.

SAFE does have messaging features built in – There is no doubt eventually there could be distributed computing built in — The question is how does this differ from current technologies?

BotNets are a dime a dozen, if that – There is a free market of them via TOR and the dark web. I don’t see SAFE as a game-changer here, aside from making our communication more secure to prevent users from accidentally becoming part of a Bot-net…

Thus you negate my condition of it happening “on its own”. Its done apart from the non connective structure proposed for the computing element of SAFE in order to make it connective which is a requirement for any AI/brain

Granted that maybe a separate computing feature could be added even later on to provide distributed computing by the network (without user equipment/software) and that may enable the connectivity required to enable in the far future some sort of self aware AI