The Nightstar Zoo

Nightstar IRC Network - irc.nightstar.net
It is currently Fri Nov 24, 2017 6:24 am

All times are UTC - 6 hours [ DST ]




Post new topic Reply to topic  [ 31 posts ]  Go to page 1, 2  Next
Author Message
PostPosted: Thu Apr 27, 2017 10:28 pm 
Offline
Safari Exhibit
Safari Exhibit

Joined: Mon Aug 13, 2012 10:03 pm
Posts: 198
It's a good thing they got a huge deposit up front. It sort of looks like the Toughs might have to collect the balance at gunpoint, while defending their employer from some sort of omniscient brain-sucking computer system.

RED-REO zombie hack mk II, anyone?

Whups, think I hit myself with a jargon zapper again. Let me try that again.

Since the immortality protocol that Petey released was supposed to be immune to getting hacked and zombified, has anyone else noticed that the ancient star-engulfing AI blew through Petey's RED-REO security patches like they weren't even there? It at least had the courtesy to back its victims up before hijacking their meat shells, but there's no reason to assume that it can't do the same brain-jacking to every meat shell that comes into communications range with it or any of its puppets. Which means the Toughs' "immortality" could be very, very short-lived.

_________________
May or may not run a Reaver-powered Perpetual Internal Combustion Engine of War And Shrapnel Fabber


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 1:31 am 
Offline
Janitor
Janitor
User avatar

Joined: Fri Jun 22, 2012 5:44 pm
Posts: 2329
dire wrote:
Since the immortality protocol that Petey released was supposed to be immune to getting hacked and zombified, has anyone else noticed that the ancient star-engulfing AI blew through Petey's RED-REO security patches like they weren't even there?

That's of course supposing this race is A) on the list of races that RED-REO will work on, B) had the resources to outfit those crews with it, and C) that their mission began after RED-REO was shared with the galaxy.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 2:29 am 
Offline
Reptile House Exhibit
Reptile House Exhibit

Joined: Sat Nov 09, 2013 1:22 pm
Posts: 226
evileeyore wrote:
That's of course supposing this race is A) on the list of races that RED-REO will work on, B) had the resources to outfit those crews with it, and C) that their mission began after RED-REO was shared with the galaxy.

Has it been shared with the galaxy as a whole? We saw it shared with the UNS, but that is the government of Sol System, not of the galaxy. Giving it to them required also giving them a vast pile of PTUs to avoid a resource crisis.

The fact that this enabled the Offans to acquire a battleplate's worth of fabbers to get civilisation restarted in the can, and killed off a hostile faction of UNS Intelligence wasn't the point at all. Honestly.

_________________
Weapon: Meal, Ready to Explode.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 5:01 am 
Offline
Janitor
Janitor
User avatar

Joined: Fri Jun 22, 2012 5:44 pm
Posts: 2329
John Dallman wrote:
Has it been shared with the galaxy as a whole?

Unlikely.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 10:15 am 
Offline
Entertainment
Entertainment

Joined: Tue Jun 12, 2012 10:54 am
Posts: 723
evileeyore wrote:
John Dallman wrote:
Has it been shared with the galaxy as a whole?

Unlikely.

It was developed and actually shared by Petey, so... Almost certainly. And if not, it's only a matter of time.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 10:34 am 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Fri Jul 01, 2016 8:10 am
Posts: 385
'hardened and unhackable' only implies the presence of 'unfeasibly'. When you have a Matryoska Brain, you're several orders of magnitude higher on the computing food chain than any encryption available to people who *don't* have one.

In times past, a simple eight character password would have taken years to crack with the entire computing power of the planet. Now you can do it at home.

There's more computing power in your phone than was used in the entire Apollo program.

Assuming quantum computing is used in its construction, there's likely more processing power in the All-Star than in the entirety of the current galactic civilization.

It can spare a few cycles to brute force it's way through the intrusion countermeasures.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 6:53 pm 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Tue Aug 26, 2014 1:53 pm
Posts: 368
There are encryption systems that are proof against quantum computing (not relying on factoring a large number).

Quote:
The fact that this enabled the Offans to acquire a battleplate's worth of fabbers to get civilisation restarted in the can, and killed off a hostile faction of UNS Intelligence wasn't the point at all. Honestly.

Except that they got shortchanged, remember? Expected 5 annies, only got three.

_________________
I hope Para never plays with Tenzy. The result would be terrifying.
Para and Petey need to have some dialog together. Just because.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 7:09 pm 
Offline
Entertainment
Entertainment

Joined: Tue Jun 12, 2012 10:54 am
Posts: 723
The only truly unhackable encryption we know of is One Time Pad. Which is... impractical, for a variety of reasons.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 8:05 pm 
Offline
Safari Exhibit
Safari Exhibit

Joined: Mon Aug 13, 2012 10:03 pm
Posts: 198
JohnSmith wrote:
The only truly unhackable encryption we know of is One Time Pad. Which is... impractical, for a variety of reasons.


Why so? Salting your hash is a pretty standard industry practice among people who like to keep their passwords un-cracked. It's possible to use a reversable dynamic salt, although it increases the complexity of the system a bit. So, say I salt my passwords with the first three letters of the username. I now have a one-time pad for every password in my database. Password "password" does not hash into the same value for user Bob as it does for user Alice, but -does- hash the same between Bob and BobTheDinosaur.

Disclaimer, this sort of falls within my industry but is not my specialty. If I wanted to implement this I'd google a tutorial by someone who knew what they were doing.

_________________
May or may not run a Reaver-powered Perpetual Internal Combustion Engine of War And Shrapnel Fabber


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 8:18 pm 
Offline
Entertainment
Entertainment

Joined: Tue Jun 12, 2012 10:54 am
Posts: 723
You misunderstand how encryption works, dire. Salting your password hashes protects you against rainbow tables. It's something you do in case your password hash database is breached. It does absolutely nothing for the actual cryptographic security. It's sure as hell not a OTP.

OTP is combining your data to be encrypted with a key at least as long as the data. It requires the key to be used only once, and then permanently destroyed. It has mathematically provable perfect security. It's also a royal pain in the arse to use, because it requires your keys to be pre-shared.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 8:32 pm 
Offline
Safari Exhibit
Safari Exhibit

Joined: Mon Aug 13, 2012 10:03 pm
Posts: 198
No, I'm pretty sure I understand how encryption works. An ordinary salt is a public element of the password and counters rainbow attacks. A dynamic key not stored anywhere but the encryption code is a secret key, and because the key is dynamic, each password effectively has its own encryption algorithm.

OTP sounds less than useful on anything bigger than a password. If you're trying to encrypt the entire contents of someone's brain, you probably don't want to store another entire brain worth of encryption garbage.

We might be confusing each other a bit because password encryption is inherently one-way. We don't need to have reversable encryption, we only need to be able to encrypt the same password the same way every time.

_________________
May or may not run a Reaver-powered Perpetual Internal Combustion Engine of War And Shrapnel Fabber


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 9:14 pm 
Offline
Entertainment
Entertainment

Joined: Fri Jul 31, 2015 10:40 pm
Posts: 619
Location: Land of the webbed feet
OTP is a pain in the backside, alright . . . and for most uses, it must be reversible. Encrypted Passwords don't actually transfer information as such, they're just a go/nogo. To transfer information, you must be able to reverse the encryption and get the original plaintext back out.

Also, AFAIK, OTP can be used to encrypt something longer than itself, but there is a limit to how much longer. If you encrypt too long of a message, the text becomes crackable through frequency analysis, though there are ways to screw with that, too.

Computers have changed the details, such as how long the OTP must be in order to adequately encipher the plaintext, but the basics remain the same.

BTW, one of the ways of screwing up frequency analysis is to include common words in your OTP . . . so that instead of enciphering the word for as the letter f, the letter o, and the letter r, you encode it as SDTVP.

--FreeFlier


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 9:35 pm 
Offline
Safari Exhibit
Safari Exhibit

Joined: Mon Aug 13, 2012 10:03 pm
Posts: 198
Yeah, OTP would be very problematic for real world use where bidirectional encryption is required. For starters, as soon as your pad is exposed, the whole thing is dead in the water. Storage and transmission of the pad is problematic. Even a relatively small pad that's eg 10% of the data size is going to be large for very large data sets -- if you're trying to encrypt a 1TB hard drive, you wind up with a 100GB encryption key.

Going back to schlock mercenary, even if the backup of a given brain is encrypted, the brain itself would generally have to run in the clear in order to execute the contents of the brain. If the Matrioshka Deity can read a brain remotely, then it can copy that brain remotely. The problem IMO comes when it decides to overwrite. We haven't seen anything that could transmit data through astronomical distances (even local astronomy), penetrate a ship's hull (and possibly shields) and hack a brain. There's no reason not to assume at this point that the Matrioshka Brain is capable of, and intends, some sort of grey goo/zombie apocalypse.

Unlike most AI's, which are connected to the hypernet, a local brain would not be, and Commodore Tagon's use of a vid screen for his call to this client suggests that local brains still are not connected to the hypernet.

Also, pass/fail is only one bit, but it's an important bit :p

_________________
May or may not run a Reaver-powered Perpetual Internal Combustion Engine of War And Shrapnel Fabber


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 9:51 pm 
Offline
Entertainment
Entertainment

Joined: Tue Jun 12, 2012 10:54 am
Posts: 723
I did say that OTP had issues. But it still stands, so far as I know, as the only encryption with 100% ironclad mathematical proof of secrecy.


Top
 Profile  
 
PostPosted: Fri Apr 28, 2017 10:20 pm 
Offline
Entertainment
Entertainment

Joined: Fri Jul 31, 2015 10:40 pm
Posts: 619
Location: Land of the webbed feet
JohnSmith wrote:
I did say that OTP had issues. But it still stands, so far as I know, as the only encryption with 100% ironclad mathematical proof of secrecy.

As long as the rules for use are followed.

When people start breaking the rules, then things go haywire.

--FreeFlier


Top
 Profile  
 
PostPosted: Sat Apr 29, 2017 1:53 am 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Tue Aug 26, 2014 1:53 pm
Posts: 368
For OTP, you can just break your OTP into three parts, via XOR. Each part can be shipped, and as long as no one but your target gets all three parts, you're fine. A single one-inch cube of high-density 3d storage can store more than enough information. All you need to worry about is some super-intelligence getting ahold of one part, deciphering what your method for generating the xor code was, and then determining the other two from that. But how often would something as incredulous as that happen? :-)

_________________
I hope Para never plays with Tenzy. The result would be terrifying.
Para and Petey need to have some dialog together. Just because.


Top
 Profile  
 
PostPosted: Sat Apr 29, 2017 11:11 am 
Offline
Aquarium Exhibit
Aquarium Exhibit

Joined: Thu Apr 20, 2017 11:51 pm
Posts: 19
Securing RED-REO isn’t going to be much like encryption security, it will be more like hacking an active computer.

If the brain copy in the skin is encrypted the hacker doesn’t care, they don’t want the data they want to control the body, so it just overwrites the brain (and backups) with the mind it wants to be in use.

The reason this can be done to somebody with early RED-REO but not a normal person is that the hack leverages the RED-REO nannies that can read/write/edit organic memory. The public release of RED-REO has been secured against the exploits that where used at oisri and any others that peaty found. Given RED-REO nannies have little need for regular communication a key defence would be to limit their available communications bandwidth and range.

Now I believe the Matrosca Brain’s attack was not mediated through RED-REO after all it had no way to know it existed less the specifics of how it could be hacked, nigh infinite processing power won’t solve a problem if you don’t have enough information to find the answers (the only people to have hacked any version of RED-REO had full access to its development code to look for access).

If all the above assumptions are correct then RED-REO may have been an effective defence against the Matrosca Brain’s mind hack. The Matrosca Brain overwrites the brain and assumes success the RED-REO detects the attack and overwrites the brain with the dermal backup of your original personality.



As to OTPs. They are extraordinarily powerful if you can distribute the pads. They are most effective in a situation where 2 people are together and with to separate but maintain secure communications. For example the navy could hand a terabyte hard drive full of one time pad to a ship captain and then be able to send him a full terabyte of orders with perfect security, this should be ample if you don’t send sensitive orders as video. Other ships have different pads. The admirals ship could also have pads paired to ships under his command, an aircraft can carry a large enough pad to secure a full missions voice communications text orders still images and a decent chunk of video if needed.

If you are trying to establish secure communications from a distance they are impractical.

There is also an issue in how to generate them. If you want mathematically perfect encryption the pad itself has to be truly random. Generating truly random numbers is hard. Computers can’t do it, humans suck at it (and are worse at recognising it) and every system that has been found that we cant crack is slow, if you want a terabyte of true random numbers your going to have to hook a computer to a measuring device and measure something that is constantly changing in a truly random fashion, and each measurement gets you only a couple of bits you will need to make and process trillions of observations for each terabite pad.


Top
 Profile  
 
PostPosted: Sat Apr 29, 2017 1:16 pm 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Fri Jul 01, 2016 8:10 am
Posts: 385
There's no such thing as *perfect* security, only 'security we don't have the power to crack yet'. So long as something is known by somebody, somewhere, there's a weak link.

To a Matryoshka Brain, the computers are the weakest link.

We're talking about a hundreds of thousands of miles thick shell of computing components surrounding a star, using at *least* quantum computing technology. A quick scan through the computers memory will tell it everything it needs to know to have a baseline, then it's just a matter of shoving computing power at the problem - which it has in spades.

You remember how Oisri had its surface fractally storing information down to the atomic scale? And just scanning it would take hundreds of years? Oisri is probably smaller that that *probe* the Brain launched. And the vast majority of the Brain is going to be processing and memory.

Kevyn once bought a pack of quantum computing components the size of a mini-fridge with some pretty impressive sounding stats. Multiply that by a sphere around the thickness of the distance between the earth and the moon.

That's assuming they don't have an even more efficient method of processing than quantum trinary.

Point being, what the Brain probably did was what Para told the Doc she did to hack the Tarbots and Bristlecone; hacked the backups and forced a system restore to implement the modified code. This would have had the benefit of giving the Brain an entirely up-do-date copy of their brains without using scanning technology that would more likely *fry* the brains it was trying to scan. The RED-REO blood nannies are more than likely already hypernet-enabled to allow for offsite backups.


Top
 Profile  
 
PostPosted: Sat Apr 29, 2017 1:28 pm 
Offline
Reptile House Exhibit
Reptile House Exhibit

Joined: Sat May 14, 2011 10:10 pm
Posts: 344
thezombiekat wrote:
Securing RED-REO isn’t going to be much like encryption security, it will be more like hacking an active computer.

If the brain copy in the skin is encrypted the hacker doesn’t care, they don’t want the data they want to control the body, so it just overwrites the brain (and backups) with the mind it wants to be in use.

The reason this can be done to somebody with early RED-REO but not a normal person is that the hack leverages the RED-REO nannies that can read/write/edit organic memory. The public release of RED-REO has been secured against the exploits that where used at oisri and any others that peaty found. Given RED-REO nannies have little need for regular communication a key defence would be to limit their available communications bandwidth and range.

Now I believe the Matrosca Brain’s attack was not mediated through RED-REO after all it had no way to know it existed less the specifics of how it could be hacked, nigh infinite processing power won’t solve a problem if you don’t have enough information to find the answers (the only people to have hacked any version of RED-REO had full access to its development code to look for access).

If all the above assumptions are correct then RED-REO may have been an effective defence against the Matrosca Brain’s mind hack. The Matrosca Brain overwrites the brain and assumes success the RED-REO detects the attack and overwrites the brain with the dermal backup of your original personality.



As to OTPs. They are extraordinarily powerful if you can distribute the pads. They are most effective in a situation where 2 people are together and with to separate but maintain secure communications. For example the navy could hand a terabyte hard drive full of one time pad to a ship captain and then be able to send him a full terabyte of orders with perfect security, this should be ample if you don’t send sensitive orders as video. Other ships have different pads. The admirals ship could also have pads paired to ships under his command, an aircraft can carry a large enough pad to secure a full missions voice communications text orders still images and a decent chunk of video if needed.

If you are trying to establish secure communications from a distance they are impractical.

There is also an issue in how to generate them. If you want mathematically perfect encryption the pad itself has to be truly random. Generating truly random numbers is hard. Computers can’t do it, humans suck at it (and are worse at recognising it) and every system that has been found that we cant crack is slow, if you want a terabyte of true random numbers your going to have to hook a computer to a measuring device and measure something that is constantly changing in a truly random fashion, and each measurement gets you only a couple of bits you will need to make and process trillions of observations for each terabite pad.


I've heard of hooking a computer up to a Geiger counter, tuned so it produces statistically 50% '1's, and 50% '0's.
I think Random.org uses a radio tuned to white noise.

Of course, this means you can defeat random.org by parking a transmitter near their receiver, broadcasting only apparently random noise, and thus poisoning their seed.
Of course, you could just as easily park your own receiver near theirs, and get the same sample.
Probably more reliable to broadcast "white noise" on a 24 hour loop. That way, you're not trying to figure out WHICH patch of almost matches white noise was used for THIS message, as, ultimately you'd only have a limited set of samples to try. Also more likely to get caught.

A rack of Geiger counters, each with its own minute sample of Americium, would probably be denser, and more reliable against outside poisoning or sampling.
You could probably even make your own Geiger Counter on a Chip. It's not like you need the mica window if you can just ramp up the sensitivity until you get 50% static, or even put the sample right in the chip.
It might even be possible to hack something together out of an old smoke detector, without taking any pieces out of it.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 1:35 am 
Offline
Aquarium Exhibit
Aquarium Exhibit

Joined: Thu Apr 20, 2017 11:51 pm
Posts: 19
Kendrakirai wrote:
There's no such thing as *perfect* security, only 'security we don't have the power to crack yet'. So long as something is known by somebody, somewhere, there's a weak link.

for encryption a random OTP is mathematically perfect. there is no amount of processing power that will decrypt it, there are of cause non-computational attacks, such as stealing a copy of the pad.

for system security, there are defenses that work against vastly more powerful attackers. the simplest is a password portal that will not let you enter a new password for some time after you enter an incorrect password.
Quote:
To a Matryoshka Brain, the computers are the weakest link.

We're talking about a hundreds of thousands of miles thick shell of computing components surrounding a star, using at *least* quantum computing technology. A quick scan through the computers memory will tell it everything it needs to know to have a baseline, then it's just a matter of shoving computing power at the problem - which it has in spades.

You remember how Oisri had its surface fractally storing information down to the atomic scale? And just scanning it would take hundreds of years? Oisri is probably smaller that that *probe* the Brain launched. And the vast majority of the Brain is going to be processing and memory.

Kevyn once bought a pack of quantum computing components the size of a mini-fridge with some pretty impressive sounding stats. Multiply that by a sphere around the thickness of the distance between the earth and the moon.

That's assuming they don't have an even more efficient method of processing than quantum trinary.

processing has an energy cost. how big of a solar panel would it take to run Kevyn's a pack of quantum computing components the size of a mini-fridge, i doubt they would fit on the side of the mini-fridge.

a Matryoshka Brain is powered from its enclosed star and while that is a lot of energy it is also spread over a lot of area and the full thickness of the shell is multiplied by that area. the available energy density to power a Matryoshka Brain is very low. it must be designed first for efficient energy utilization not efficient volume utilization, still more processing power than anybody else has but not as much as you thought.
Quote:
Point being, what the Brain probably did was what Para told the Doc she did to hack the Tarbots and Bristlecone; hacked the backups and forced a system restore to implement the modified code. This would have had the benefit of giving the Brain an entirely up-do-date copy of their brains without using scanning technology that would more likely *fry* the brains it was trying to scan. The RED-REO blood nannies are more than likely already hypernet-enabled to allow for offsite backups.

but RED-REO blood nannies are not going to be hypernet-enabled.

the planet mercenary RPG lists remote backup as not possible but there are rumors some people have the capability (the comic shows us the toughs do have it, the combination tells us it is not part of the widely distributed system).

even the toughs offsite backups can not be updated or restored remotely. recall that when Ebby was shot through the pelvic cradle (that being where uniocs keep their brain) the doctor had to insert a physical communications device into the wound to establish contact with the nanits.

this is an example of good security for an important system. if nothing can communicate with it it can not be hacked, so if it is important and constant communication is not required, don't implement a capability for constant communication.

also, nannies are tiny, no space for a hypernode, but they could use other coms systems.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 1:52 am 
Offline
Aquarium Exhibit
Aquarium Exhibit

Joined: Thu Apr 20, 2017 11:51 pm
Posts: 19
Sean wrote:
thezombiekat wrote:
snip


I've heard of hooking a computer up to a Geiger counter, tuned so it produces statistically 50% '1's, and 50% '0's.
I think Random.org uses a radio tuned to white noise.

Of course, this means you can defeat random.org by parking a transmitter near their receiver, broadcasting only apparently random noise, and thus poisoning their seed.
Of course, you could just as easily park your own receiver near theirs, and get the same sample.
Probably more reliable to broadcast "white noise" on a 24 hour loop. That way, you're not trying to figure out WHICH patch of almost matches white noise was used for THIS message, as, ultimately you'd only have a limited set of samples to try. Also more likely to get caught.

A rack of Geiger counters, each with its own minute sample of Americium, would probably be denser, and more reliable against outside poisoning or sampling.
You could probably even make your own Geiger Counter on a Chip. It's not like you need the mica window if you can just ramp up the sensitivity until you get 50% static, or even put the sample right in the chip.
It might even be possible to hack something together out of an old smoke detector, without taking any pieces out of it.


that is basically how it would be done. although if well-implemented poisoning the seed would be very difficult, only the least significant bits of the measured quality would be used, if you try to poison the seed you are only adding to the signal, if you don't know what it was before you added to it you can't control the final number and even a small physical separation or tiny differences in hardware will cause small variations.

but it is still slow to generate, in 19 years random.org has delivered only 1.47 trillion random bits less than one-fifth of the terabyte I wanted to give to each ship in my fleet. if you actually want to implement OTPs for secure communications on the scale that military's and banks want to communicate you're going to need huge banks of sensors collecting random bits.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 5:11 pm 
Offline
Janitor
Janitor
User avatar

Joined: Sun Apr 11, 2004 10:28 pm
Posts: 1951
Kendrakirai wrote:
There's more computing power in your phone than was used in the entire Apollo program.


I keep hearing that myth. Don't confuse the Apollo Guidance Computer with "the entire Apollo Program." TYou have no idea what a Sigma Nine could do, and the MSC had a lot of them.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 5:17 pm 
Offline
Safari Exhibit
Safari Exhibit

Joined: Mon Aug 13, 2012 10:03 pm
Posts: 198
thezombiekat wrote:
...

but it is still slow to generate, in 19 years random.org has delivered only 1.47 trillion random bits less than one-fifth of the terabyte I wanted to give to each ship in my fleet. if you actually want to implement OTPs for secure communications on the scale that military's and banks want to communicate you're going to need huge banks of sensors collecting random bits.


Or else a medium-sized bank full of very small sensors.

I'm curious though, what's the difference between using a series of random seeds to power a pseudorandom RNG up to the point where it starts repeating itself, vs using a new true-random value for each element? Wouldn't that be the same thing in terms of cryptographic security? I mean, if you are confident that an RNG's seed can't be derived given X iterations of the RNG, then you can use the RNG to divide the number of necessary true random seeds by X. Use two separate RNG functions to divide the number of needed true-random seeds by 2X. And so on.

Also, why is using the microsecond timestamp to seed an RNG not true-random? Wouldn't that be "good enough" for most OTP applications?

_________________
May or may not run a Reaver-powered Perpetual Internal Combustion Engine of War And Shrapnel Fabber


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 6:07 pm 
Offline
Reptile House Exhibit
Reptile House Exhibit

Joined: Sat May 14, 2011 10:10 pm
Posts: 344
ChowGuy wrote:
Kendrakirai wrote:
There's more computing power in your phone than was used in the entire Apollo program.


I keep hearing that myth. Don't confuse the Apollo Guidance Computer with "the entire Apollo Program." TYou have no idea what a Sigma Nine could do, and the MSC had a lot of them.


Let's just look at it from the perspective of Moore's Law, but let's start somewhere convenient.
My cell phone is probably about as powerful as an old Intel Pentium III computer. Pentium III's were cutting edge around the turn of the millennium; 17 years ago.

Moores Law, since some time in the early 80's, is something like a doubling of circuit density every two years.
For convenience sake, let's say that's 20 years, so we can just say 79-99, (although 79, 80, probably 81 and 82, were still doubling annually.) This means a one-thousand fold increase in computer chip density, for a given size package, from 79 to 99.
This means that your phone is roughly one thousand times more powerful than the best MICRO-computers available in the late 70's.

Of course, we're talking about maybe a MINI-computer, or more likely a mainframe, not a micro computer.
Still, from 69-79, we can probably assume an increase in capacity on par with Moore's Law:the early years. That is, a doubling every year, roughly.
That would make your phone about a million times denser than a theoretical 1969 micro-computer, and probably still several thousand times denser, and probably more powerful, than the room-sized mainframe computers then available.

In fact, just to google it.
https://en.wikipedia.org/wiki/SDS_Sigma_series
Gives a value of 3.8us per 32 bit instruction for 'a' Sigma 9. (Same brand? Apparently.)
https://books.google.com/books?id=_itgx ... nd&f=false
gives us a value of 800,000 instructions per second.
Let's use the second, larger, value.

I didn't find a quick answer for the instructions per second of a cellphone, so I substituted a Raspberry Pi, which uses the same chip as many common smart phones.
http://hackaday.com/2016/03/01/pi-3-ben ... e-is-true/
Just using the first value, which is actually Instructions Per Second, and limiting myself to the second generation of the unit, I get 1,822, million, instructions per second.

1,822,000,000/800,000=2,277.5 times the capacity.

So, unless the Apollo project had over two thousand Sigma 9's in service, yes, there is more power in your phone than in all the digital computers combined, that were used to successfully execute the Apollo Program.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 6:14 pm 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Fri Jul 01, 2016 8:10 am
Posts: 385
Sean wrote:
ChowGuy wrote:
I keep hearing that myth. Don't confuse the Apollo Guidance Computer with "the entire Apollo Program." TYou have no idea what a Sigma Nine could do, and the MSC had a lot of them.


Let's just look at it from the perspective of Moore's Law, but let's start somewhere convenient.
My cell phone is probably about as powerful as an old Intel Pentium III computer. Pentium III's were cutting edge around the turn of the millennium; 17 years ago.

Moores Law, since some time in the early 80's, is something like a doubling of circuit density every two years.
For convenience sake, let's say that's 20 years, so we can just say 79-99, (although 79, 80, probably 81 and 82, were still doubling annually.) This means a one-thousand fold increase in computer chip density, for a given size package, from 79 to 99.
This means that your phone is roughly one thousand times more powerful than the best MICRO-computers available in the late 70's.

Of course, we're talking about maybe a MINI-computer, or more likely a mainframe, not a micro computer.
Still, from 69-79, we can probably assume an increase in capacity on par with Moore's Law:the early years. That is, a doubling every year, roughly.
That would make your phone about a million times denser than a theoretical 1969 micro-computer, and probably still several thousand times denser, and probably more powerful, than the room-sized mainframe computers then available.

In fact, just to google it.
https://en.wikipedia.org/wiki/SDS_Sigma_series
Gives a value of 3.8us per 32 bit instruction for 'a' Sigma 9. (Same brand? Apparently.)
https://books.google.com/books?id=_itgx ... nd&f=false
gives us a value of 800,000 instructions per second.
Let's use the second, larger, value.

I didn't find a quick answer for the instructions per second of a cellphone, so I substituted a Raspberry Pi, which uses the same chip as many common smart phones.
http://hackaday.com/2016/03/01/pi-3-ben ... e-is-true/
Just using the first value, which is actually Instructions Per Second, and limiting myself to the second generation of the unit, I get 1,822, million, instructions per second.

1,822,000,000/800,000=2,277.5 times the capacity.

So, unless the Apollo project had over two thousand Sigma 9's in service, yes, there is more power in your phone than in all the digital computers combined, that were used to successfully execute the Apollo Program.


I'd just like to add that, even something much more recent, the Hubble Telescope, uses a 486 processor. I think that's even in its upgrade unit that it needed to correct the flaw that made it nearsighted.

According to that wikipedia article linked, at the end of its life, the Sigma-9 series was discontinued because the newer computers could *emulate* it - which implies a greater enough jump in processing speed that the overhead of the emulation process was matched or overcome by the newer hardware.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 9:31 pm 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Tue Aug 26, 2014 1:53 pm
Posts: 368
dire wrote:
thezombiekat wrote:
...

but it is still slow to generate, in 19 years random.org has delivered only 1.47 trillion random bits less than one-fifth of the terabyte I wanted to give to each ship in my fleet. if you actually want to implement OTPs for secure communications on the scale that military's and banks want to communicate you're going to need huge banks of sensors collecting random bits.


Or else a medium-sized bank full of very small sensors.

I'm curious though, what's the difference between using a series of random seeds to power a pseudorandom RNG up to the point where it starts repeating itself, vs using a new true-random value for each element? Wouldn't that be the same thing in terms of cryptographic security? I mean, if you are confident that an RNG's seed can't be derived given X iterations of the RNG, then you can use the RNG to divide the number of necessary true random seeds by X. Use two separate RNG functions to divide the number of needed true-random seeds by 2X. And so on.


First, there's a difference between "We've only distributed 1.47 trillion random bits" and "we've only generated 1.47 trillion random bits". Any time that their buffer is full, and data is generated, it is wasted.

Random number generators that take the seed as "occasional re-input to add more entropy to the sequence" do exist; it is basically the next generation of RNG improvement. The idea is that by the time you have seen enough bits of random data to have figured out the content of the first pool of random data, some of the second pool has been used to alter the first pool. Now you have to re-determine the first pool's content from scratch, and by the time you've done that, more of the second pool has polluted the first.

Eventually you've seen enough that you can decipher not only what was in the first pool, and the second. But before you can now say "I know what will come next", there's a third pool that has now inflicted itself on the rest.

But note the different goals here. This is about making it impossible to tell what will come next out of the RN generator. Which is not the same as trying to decipher that the RNG spat out in the past.

Even if I can't figure out what will come next, I might be able to tell what came before. And if I can figure out what came before, then I can re-create the OPT that you generated in the past.


Quote:
Also, why is using the microsecond timestamp to seed an RNG not true-random? Wouldn't that be "good enough" for most OTP applications?


Geez, if I know approximately when you ran it, then I've got a good start on breaking your random sequence. That's before taking into account that a microsecond timestamp has very few effective bits.

_________________
I hope Para never plays with Tenzy. The result would be terrifying.
Para and Petey need to have some dialog together. Just because.


Top
 Profile  
 
PostPosted: Sun Apr 30, 2017 9:51 pm 
Offline
Intern
Intern
User avatar

Joined: Mon Jun 29, 2015 9:52 pm
Posts: 1088
Well since we're going down this train, anyone heard of nomx (NOT Nomex)?
https://arstechnica.com/information-tec ... -protocol/

_________________
If you need to use a squeegee, they're probably at least Laz-4


Top
 Profile  
 
PostPosted: Mon May 01, 2017 1:10 am 
Offline
Aquarium Exhibit
Aquarium Exhibit

Joined: Thu Apr 20, 2017 11:51 pm
Posts: 19
dire wrote:
thezombiekat wrote:
...

but it is still slow to generate, in 19 years random.org has delivered only 1.47 trillion random bits less than one-fifth of the terabyte I wanted to give to each ship in my fleet. if you actually want to implement OTPs for secure communications on the scale that military's and banks want to communicate you're going to need huge banks of sensors collecting random bits.


Or else a medium-sized bank full of very small sensors.

actually, you need a very large bank of sensors. they need to be sensitive enough to fluctuate with low-level noise, so more sensitive than the pixels in most digital cameras, and 10 observations per sensor per second is reasonable. assuming you want to generate a terabyte in a year. you need about 350 high precision sensors assuming no down time. to produce just one pad per year.
Quote:
I'm curious though, what's the difference between using a series of random seeds to power a pseudorandom RNG up to the point where it starts repeating itself, vs using a new true-random value for each element? Wouldn't that be the same thing in terms of cryptographic security? I mean, if you are confident that an RNG's seed can't be derived given X iterations of the RNG, then you can use the RNG to divide the number of necessary true random seeds by X. Use two separate RNG functions to divide the number of needed true-random seeds by 2X. And so on.

you can. but there is now a mathematical relationship between elements of the pad. could it be cracked, almost certainly not, but that is true if you just use a really long encryption key. if your paranoid enough to use an OTP you probably have your heart set on mathematical perfection wich means no shortcut.
Quote:
Also, why is using the microsecond timestamp to seed an RNG not true-random? Wouldn't that be "good enough" for most OTP applications?

now, this is a really bad idea.

for generating normal encryption keys it kind of works, although if you know approximately when it was run it is possible to guess parts of the key.

if you generate an OTP you are generating huge numbers of random bytes in rapid succession, with a modern computer several digits will be collected every microsecond, if you discard all but the last digit (so an attacker can't use knowing the year it was generated as a shortcut) your OTP winds up looking a lot like this
1111122222233333333344444555555666666667777778888899999900000000111111122222233333334444444
you can manipulate that a bit but in the end, there will be a mathematical relationship between the OTP and a series of repeating numbers incrementing at intervals that are moderately predictable.


Top
 Profile  
 
PostPosted: Mon May 01, 2017 4:27 am 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Tue Jun 26, 2012 5:29 am
Posts: 524
Alternately, the very smart brain decodes the teraport authentication of the ship, teraports brainjacker probes on top of everyone's head, and the brainjacker probes physically remove the meatform's brains and replace it with their own organic substrate that superficially mimics the intelligences they have just cut out of its skull by means of their previous scans of said brains.

Lots can be done with physical access. All it needs is that.


Top
 Profile  
 
PostPosted: Mon May 01, 2017 8:10 am 
Offline
Monkey House Exhibit
Monkey House Exhibit

Joined: Fri Jul 01, 2016 8:10 am
Posts: 385
Reaver225 wrote:
Alternately, the very smart brain decodes the teraport authentication of the ship, teraports brainjacker probes on top of everyone's head, and the brainjacker probes physically remove the meatform's brains and replace it with their own organic substrate that superficially mimics the intelligences they have just cut out of its skull by means of their previous scans of said brains.

Lots can be done with physical access. All it needs is that.


It probably wouldn't have been quite so up to date with everyone's memories then, since that one guy was able to watch a missile come right for him through the window.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 31 posts ]  Go to page 1, 2  Next

All times are UTC - 6 hours [ DST ]


Who is online

Users browsing this forum: Bing [Bot], Google [Bot] and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group