Hacking 2000 years from now...

Isn't going to be possible. See:

https://www.wired.com/2016/09/computer- ... roof-code/

There are things we give SF authors without a quibble, like faster than light space drives. However, we raise a bit of a fuss when predictable developments in current technologies seem not to have happened because plot.

Software development is my field. I've known about formal methods since the 80s. It's nice to see that they're advanced enough to get out of the toy system stage into actual real-world application. In another 2000 years I don't think that the word "hacker" will exist except in dictionaries of obsolete words. Heck, I doubt if it'll exist in 100 years.

JohnRoth wrote:Isn't going to be possible. See:

https://www.wired.com/2016/09/computer- ... roof-code/

Key parts of Little Bird’s computer system were unhackable with existing technology, its code as trustworthy as a mathematical proof.

Like the Armor vs Weapons balance, "this too shall pass." Technology to hack the unhackable is going to be developed and a new way to make systems "unhackable" will be developed -- lather, rinse, repeat, ad nauseum.

Cowboys say "Aint't a horse that cain't be rode, ain't a rider that cain't be throwed".
I do think that weapons are now ahead of armour and will stay there, but I'm not sure of hacking.
John Ringo had a series thst featured an android killer rabbit that had been developed to rid the world of spammers. So efficient that people didn't know what spam was anymore.

Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

cthia wrote:Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

And even with mathematically provable security you are still at risk if something that was taken as axiomatic turns out not to be universally true. Even something as simple and fundamental as the CPU correctly running the instructions provided to it isn't guaranteed in the real world in the face of fault injection attacks. Of and you can't discount operator stupidity; the operators have to have some way to authenticate themselves to the system and lazy people are endlessly inventive at finding ways to make their computer interactions easier even at the expense of security. Anything from writing down passwords near the console to leaving the smart card with the certificate credentials under the smart card reader, or in the nearby desk drawer. If there's anyway for an operator to undermine the security some operator will.

cthia wrote:Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

Old joke:

Patient: Doctor, it hurts when I do this.

Doctor: Don't do that.

There are certain statements that I regard as suggestive that the person making them doesn't understand the problem.

One is a reference to Godel's Incompleteness Therom and the inability to solve the Halting Problem for certain cases.

The solution is very simple: don't write programs that you can't prove correct.

The objective isn't to try to prove the correctness of software which was constructed by Klingon Software Development and which looks like the illegitimate offspring of Cthulhu and the Flying Spaghetti Monster. It's to create software that works and is bulletproof. That may require throwing what we've got out and starting over.

By the way: I regard LISP as one of the major reasons why the original approaches to artificial intelligence failed. (It's not the only one, or even the biggest.) Revisiting some of that with the advantage of another 45 years of development in linguistics is somewhere on my bucket list. It will not be in LISP.

Jonathan_S wrote:

cthia wrote:Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

And even with mathematically provable security you are still at risk if something that was taken as axiomatic turns out not to be universally true. Even something as simple and fundamental as the CPU correctly running the instructions provided to it isn't guaranteed in the real world in the face of fault injection attacks. Of and you can't discount operator stupidity; the operators have to have some way to authenticate themselves to the system and lazy people are endlessly inventive at finding ways to make their computer interactions easier even at the expense of security. Anything from writing down passwords near the console to leaving the smart card with the certificate credentials under the smart card reader, or in the nearby desk drawer. If there's anyway for an operator to undermine the security some operator will.

Most of these things are due to not paying proper attention to security in the first place. Let's take one example: why do people write down their passwords on sticky notes where someone else can find them?

There are two reasons. One is the system insists on generating strong passwords that are impossible for ordinary mortals to remember. Especially when someone may have dozens of these things. The other is that the people who actually need access in an emergency (and there is always an emergency) don't have their own, individually authenticated access.

Both of these can be chalked up to stupidity, but it's not the operators who are being stupid. It's the managers.

cthia wrote:Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

JohnRoth wrote:Old joke:

Patient: Doctor, it hurts when I do this.

Doctor: Don't do that.

There are certain statements that I regard as suggestive that the person making them doesn't understand the problem.

I was invited to the symposium, strong-armed by a colleague actually, because I do understand the problem. And because my algorithm skills were to be requested.

One is a reference to Godel's Incompleteness Therom and the inability to solve the Halting Problem for certain cases.

Not for certain cases, but for all cases that are not finite solutions with a known resolution.

The solution is very simple: don't write programs that you can't prove correct.

The problem is to write programs that the programming can prove correct. Impossible to do in such a complex system being attempted. That's why the strategy is to attack it in pieces -- to prove parts of the system foolproof. Akin to the area under the curve solution and mapping.

The objective isn't to try to prove the correctness of software which was constructed by Klingon Software Development and which looks like the illegitimate offspring of Cthulhu and the Flying Spaghetti Monster. It's to create software that works and is bulletproof. That may require throwing what we've got out and starting over.

I agree, as does most others in the field. (There was a heated debate on that at the symposium.) Yet, start over with what, so that the inherent problems of the beast won't still be there?

By the way: I regard LISP as one of the major reasons why the original approaches to artificial intelligence failed. (It's not the only one, or even the biggest.) Revisiting some of that with the advantage of another 45 years of development in linguistics is somewhere on my bucket list. It will not be in LISP.

Suit yourself.

Huh? LISP is not why the original approaches to AI failed. LISP and the lambda was what originally made it possible. The lack of hardware that could run such a powerful language on at the time and to digest such am ambitious project was the failure. This was 1958! LISP MACHINES was a start. But, the hardware and tech at the time was horribly lagging behind a language that hasn't changed since its conception, because it is so fundamentally and intrinsically sound that it doesn't have too. You don't change lambda. Lambda changes you.

Hardware and tech is ready for LISP now, and the power of LISP is still there. Unfortunately, so is the immobilizing fear of parenthesis.

You will not be able to hide from Lisps data structures of "lists." Even in the description of the goals, lists are used. Unavoidable. Unless there's a miracle language waiting in the wings.

Say what you may. Godel and the halting problem is like an awaiting croc with its mouth open -- if a truly secure, and not somewhat secure system is the goal. You can bank on it, or leave your currency under a mattress.

By the way, I recall that the halting problem was discussed at the symposium. I had originally threw it in the face of my colleague when I was invited and we discussed it beforehand. Guess who wasn't surprised when it came up at the symposium? If anyone knows anything about computers, you'd darn well know that it's a stumbling block. And a mountain of a stumbling block in this project.

Just a quick search, cross referencing HACMS with halting problem yields...

snip

Fisher became fascinated by this so-called “Halting Problem” as soon as she heard about it, in an introduction to programming class at Stanford. “The fact that you can prove something is impossible is such an amazing thing that I wanted to learn more about that domain. That’s actually why I became a computer scientist,” she says. The instructor for the class was a guy named Steve Fisher. She was interested enough in him that she wound up marrying him after school, and taking his last name.

snip

from https://www.wired.com/2012/12/darpa-drones/

Edit: Referenced Symposium...
http://cmns.umd.edu/news-events/events/1050

Jonathan_S wrote:

cthia wrote:Interesting field and the theory is quite sound but too much has to happen before it can ever become a reality. The very concepts that make it possible are the very same concepts that denies its absolute probability now — mathematical algorithms (the essence of proofs). You can base security on the superiority of certain algorithms. That has always been the case. NIST in a nutshell, for example. http://csrc.nist.gov/groups/ST/toolkit/examples.html

HACMS is taking it to a whole new level, which again is theoretically sound - if certain inroads can be made. Inroads which seem far away from being paved. Incidentally, I suggested at the symposium in certain discussions that LISP is essential and should be considered rather than Cog and Isabelle. LISP has a bigger inherent hammer and any tool missing can be easily constructed. I'd consider using NUPRL.

But at any rate there are stumbling blocks plaguing the fruition of a foolproof system that seems to me to be insurmountable, at present. Before the realization of an "absolutely" sound system, this one stumbling block still awaits up ahead...

Gödel's Incompleteness Theorems and the "Halting Problem."

One other stumbling block are current proofs which are based on known mathematical equations. There are holes in our absolute mathematical knowledge. If we had a TOE (Theory of Everything) then we'd be there, because at least these dangling, yet very important equations, (whose hopefully inevitable solutions) wouldn't be left to bite us in the arse as they add or detract from our "logic" - equations whose reminder thereof, I just posted in Free Range. Talk about timing. https://en.wikipedia.org/wiki/Millennium_Prize_Problems

A system without first solving these issues can not be considered an absolute system because of the nature of the beast. Good for a temporary system only -- which is pretty much what we have now (Which would probably be at the mercy of borderline savants who see mathematically in an abstract form) - therefore, in essence, not a reality. Yet, again, it is theoretically sound and certainly possible to use it to strengthen the lock on the door.

I attended a symposium on this a few years ago at the University of Maryland.

And even with mathematically provable security you are still at risk if something that was taken as axiomatic turns out not to be universally true. Even something as simple and fundamental as the CPU correctly running the instructions provided to it isn't guaranteed in the real world in the face of fault injection attacks. Of and you can't discount operator stupidity; the operators have to have some way to authenticate themselves to the system and lazy people are endlessly inventive at finding ways to make their computer interactions easier even at the expense of security. Anything from writing down passwords near the console to leaving the smart card with the certificate credentials under the smart card reader, or in the nearby desk drawer. If there's anyway for an operator to undermine the security some operator will.

My bold to call your attention.

Which is pretty much what I was alluding to — with the inevitable resolutions of the "Millenium Problems" (for starters) — a few clicks upstream.