Inspired by this thread. 
I’m not talking about a poorly written program that does what the writer didn’t intend it to do, but bits of random coding interacting to form computer virus.
I don’t know much about programming, so if this is a stupid question, let me know.
A poorly-written program could crash the system, fill up the disk, cause data loss, etc., but it is not going to replicate itself by deliberately infecting other programs and executable files. There is no evolutionary process or natural selection going on to produce such programs. Computer viruses are written by hand by hackers and researchers and also by criminals.
Can thread viruses on a message board form by random combinations of bits of decaying threads?
In theory, yes.
The data on a hard disk could get corrupted by a power failure in such a way that it created a new, functional program.
In practice, though, the chance of that happening is roughly the same as the chance that you will win the lottery one million times in a row.
Yeah, it’s possible in the same sense that a roomful of monkeys banging on typewriter keyboards will replicate a Shakespearean sonnet. OK, probably a bit more likely than that, since a self-replicating code snippet could be pretty short, but still very farfetched.
Or that a bomb going off in your pantry would create a dish worth of a Michelin star.
Moderator Note
It’s misplaced in General Questions, since you’ve been around long enough to know that parody threads belong in the Pit. This is closed. Don’t do this again.
Colibri
General Questions Moderator
After a PM from the OP, I have reopened the thread. It appeared to be suggesting that the question it was based on was stupid.
I can see why someone might ask this, but unless someone had programmed the code to behave in this manner, no, this could not happen.
ETA: there may be a question of semantics at work here. When you say “virus”, are you using as narrowly technically defined, or as the broader general usage definition?
What is the difference?
I am no expert on this so sorry if this is impossibly stupid, but aren’t some “neural networks” capable of modifying their own code and generating new code? Could a computer virus arise as an unintended consequence?
I don’t know what bits of random coding are. But worms have been released into the wild with unintended consequences. See the Morris worm.
The general term encompassing viruses, worms and other programmed nasties is malware.
Malware is probably what I am asking about, then.
One interesting example of a simple system with reproducing-like behavior is Conway’s Game of Life.
Early on, a “glider gun” was discovered by accident (not truly random). A pattern that repeatedly sent out “gliders” over and over. So a sort-of-randomish generation pattern was possible.
It took some time before truly self-replicating patterns were found, but not randomly at all.
OTOH, it wouldn’t take much for a very large “universe” seeded with random patterns to come up with one of these and then you would have a virus (a.k.a. “life”*).
Note that the simplicity of the rules of Life work in it’s favor. Regular computer code has much more complexity so randomly generating some code that does something interesting is much, much harder. Nevermind full self-replication.
- Conway’s naming implied such patterns presumably existed.
In 1959 a researcher from IBM did an experiment where he mutated machine language code to see if he could get a new working program. I think it was published in either the Spring Joint or Fall Joint Computing Conference (SJCC, FJCC). He ran lots of permutations, and I think he might have gotten one kind of program out of it, but it was nothing as complex as a piece of malware would be.
If you started with a working piece of malware you might accidentally produce another working piece of malware, but in general it is easier to be wrong than to be right. Didn’t some virus not activate because of a bug some time?
So the answer is not bloody likely.
In a former job a good part of my activity was devoted to scanning for malware signatures/patterns. This was on an Army contract, so the policy was enforced strictly, at least in our facility.
It was usually a good thing to find a known threat and not allow it to get onto our networks. However, the failure of “one size fits all” solutions became evident with one incident.
Our network consisted of VAX/VMS minicomputers, a large collection of different UNIX variants, Intel and AMD based PCs, and Macintoshes. But the only approved malware scanner only looked for malware aimed at Windows executables. More, every file to be imported (via diskette, tape, etc.) to any of our systems was required to pass the scanning software.
Now, one day a new release of production code came in. It was written to run in our IBM POWER6 systems using the IBM AIX operating system. The production code was written in C, C++, Ada, FORTRAN 90, PERL, and a whole lot of Bourne and Korn shell scripts. Guess what happened…
The scanner found a signature matching some malware. That malware could only execute in the Windows2000 operating environment as malware. That malware could not execute in the IBM AIX environment as malware. The production code had to be rejected for use in our facility.
(BTW, I do not fault the Army policy makers for this, as the standard was to ensure against any and all malware installations. The fact that the detected signature could not infect the AIX systems was in itself no guarantee that the code could not have been repackaged and then transmitted to a PC, where it might have been executed. I do fault the developers for ignoring their requirement to run the same scanner against their production code before it was delivered to the Army.)
I left the job shortly after that, and have no idea if the developers were able to overcome the (non-)difficulty.
Unlike organic matter, random code doesn’t collide and combine all that much. To be sure random numbers are generated all the time for simulation exercises. But those aren’t commands: they are data and interpreted as such. It isn’t hard to write a program that will create random code then attempt to execute it. But that’s not exactly a common (or rewarding) practice.
But is code written that has unintended consequences? Sure: most programming time is devoted to debugging.
No. Artificial neural networks are networks of artificial neurons, and have no “code” to modify. Neural networks are therefore less relevant to the subject of computer viruses than the traditional model of machine code executing on a von Neumann architecture.
In real life, viruses evolve as a result of biological mechanisms like selection pressure, mutation, recombination, etc. None of these apply to a computer operating system. A buggy computer program will often just crash, and the operating system is supposed to prevent it from being able to corrupt its own code or that of other programs. A piece of code that replicates itself by bypassing system permissions in order to infect other programs can only be the result of clever programming, not random bugs.
There is such a thing as evolutionary programming. Again, though, this is a deliberate type of programming via optimization, and is not suitable for developing computer malware, though if you know of such a thing it sounds quite revolutionary.