KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU (11/03/86)
From: "Keith F. Lynch" <KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU>
From: pyramid!decwrl!amdcad!lll-crg!hoptoad!tim@Ucb-Vax.arpa (Tim Maroney)
... biasing a random number generator, which is presumably a
microscopic silicon device whose functioning requires special
training for understanding.
A true random number generator might be a speck of radium with a
Geiger counter next to it. A silicon device, i.e. computer, can only
be a ~rpseudo-random number generator. The numbers that are produced by
it may be well distributed, but they only APPEAR random - because of
our lack of knowledge of the algorithm or lack of ability to compute
its results in our heads. If someone appears to demonstrate an ability
to bias a properly operating pseudo-random number generator, that is
excellent evidence that the experiment IS bogus.
...Keithzdenek@heathcliff.columbia.edu (Zdenek Radouch) (11/06/86)
In article <236@sri-arpa.ARPA> KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU writes: >From: "Keith F. Lynch" <KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU> > > A true random number generator might be a speck of radium with a >Geiger counter next to it. A silicon device, i.e. computer, can only >be a ~rpseudo-random number generator. The numbers that are produced by >it may be well distributed, but they only APPEAR random - because of >our lack of knowledge of the algorithm.... Sorry, the life isn't that simple. Let's not argue about true or pseudo random numbers, philosophers will do it for us. We use computers because it's easy to get any distribution we want and the process is controlled. If you don't understand the mechanism of generating the numbers you just get some numbers and you have to analyse them. As a result you could say they have such and such distribution, but that doesn't imply that the numbers to be generated next will have the same property. As far as the examples you used, a Geiger counter counts 0,1,2,3,4,5,6.... That sequence can hardly be considered to be random. The only way to make it what you call "true random number generator" would be to build some fairly sophisticated hardware around it. Even then it wouldn't generate the numbers with required distribution; it's simply too complicated. It would be much simpler to use an analog silicon device in a mode where "you lack the knowledge...", but this method (it is actually used) is also hard to control. A mathematical model is just the easiest and most reliable approach. zdenek ------------------------------------------------------------------------- Men are four: He who knows and knows that he knows, he is wise - follow him; He who knows and knows not that he knows, he is asleep - wake him; He who knows not and knows that he knows not, he is simple - teach him; He who knows not and knows not that he knows not, he is a fool - shun him! zdenek@CS.COLUMBIA.EDU or ...!seismo!columbia!cs!zdenek Zdenek Radouch, 457 Computer Science, Columbia University, 500 West 120th St., New York, NY 10027
cooper@pbsvax.dec.com (TOPHER COOPER) (11/08/86)
Keith F. Lynch <KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU> writes: > > From: pyramid!decwrl!amdcad!lll-crg!hoptoad!tim@Ucb-Vax.arpa (Tim Maroney) > > ... biasing a random number generator, which is presumably a > microscopic silicon device whose functioning requires special > training for understanding. > > A true random number generator might be a speck of radium with a >Geiger counter next to it. A silicon device, i.e. computer, can only >be a ~rpseudo-random number generator. The numbers that are produced by >it may be well distributed, but they only APPEAR random - because of >our lack of knowledge of the algorithm or lack of ability to compute >its results in our heads. If someone appears to demonstrate an ability >to bias a properly operating pseudo-random number generator, that is >excellent evidence that the experiment IS bogus. True random number generators are used in two ways in parapsychology. They are used in PK experiments, where the subject is instructed to "try" to get the generator to match a particular target or sequence of targets. They are also used in precognition experiments where the subject attempts to guess what the RNG will later generate. It is fairly clear that these "precognition" experiments could actually be explained by PK. Whether *all* apparent precognition can be explained in terms of PK and/or clairvoyance is a matter of serious debate in parapsychology. Spontaneous case material is highly suggestive that precognition as a distinct phenomena exists, but spontaneous case material, because of intrinsically poor controls, can never be more than suggestive. More elaborate experiments are also suggestive of "true" precognition" but again, no more than that. Anyway, RNG's come in two basic designs. Both base their randomness on a random event generator (REG). In the more common design, a high-speed toggle flip-flop (for binary "numbers") or counter is driven by a clock (generally in the MHz range). The REG generates "events" at a relatively low rate (hundredths of a second or slower). When an event is registered the clock is stopped and the counter is read as the result. The other design is in the sense the opposite. A high-speed REG "toggles" a flip-flop or counter over a fixed period of time. When the time expires, the counter is read for the result. The time is set so that the average number of events is several hundred or even thousand times the period of the counter. There are two principal designs of REGs in use as well. One, the Schmidt REG, is basically the design mentioned by Keith: a radioactive source and a Geiger counter. The other is, fundamentally, a "silicon device." In particular it is a reverse biased diode junction (usually a Zener diode). This is electronically noisy. The noise is amplified and passed through a threshold switch. The threshold is adjusted to produce the desired rate of events. The nature of the noise in semiconductor junctions is an open issue currently under investigation. The two major sources are thermal noise (which is what it sounds like) and "1/f noise" (which is a rough description of its spectrum). The current theories on both of these sources seem to agree that they both are quantum mechanically non-deterministic, i.e., that their behavior is dominated by effects at the quantum level. A third design for a REG is sometimes used for convenience but is considered much less rigorous. A human operator hits a button which stops a clock. The accuracy needed to "control" this situation is measured in milli- or microseconds, which is thought to be much smaller than the capabilities of the human nervous system. The random variability used here is that of the nervous system. This is usually used in the form of hitting a keyboard key which causes the low order bit(s) of the system clock to be read on a computer. This has the advantage of requiring no special equipment but raises a host of issues as to its reliability, theoretical correctness and interpretation. Pseudo-random number generators (PRNGs) *are* used in parapsychology; either by themselves in clairvoyance experiments or with true REGs in PK and precognition experiments. I have never heard of a case in which a pseudo-random sequence was claimed to have been "tampered with" via PK. The seed for a PRNG is always the result, however, of a mechanical process and there is strong evidence that its selection may be influenced by unconscious psi (whatever that is). The accepted practice in clairvoyance experiments is therefore to use a single very-long sequence for all experiments at a particular lab over an indefinite time period. Topher Cooper USENET: ...{allegra,decvax,ihnp4,ucbvax}!decwrl!pbsvax.dec.com!cooper INTERNET: cooper%pbsvax.DEC@decwrl.dec.com Disclaimer: This contains my own opinions, and I am solely responsible for them.
daveh@cbmvax.commodore.COM (Dave Haynie) (11/10/86)
> A third design for a REG is sometimes used for convenience but is > considered much less rigorous. A human operator hits a button which stops > a clock. The accuracy needed to "control" this situation is measured in > milli- or microseconds, which is thought to be much smaller than the > capabilities of the human nervous system. The random variability used here > is that of the nervous system. This is usually used in the form of hitting > a keyboard key which causes the low order bit(s) of the system clock to be > read on a computer. This has the advantage of requiring no special > equipment but raises a host of issues as to its reliability, theoretical > correctness and interpretation. I used something related to this for a design project in an EE class long ago. The objective of the project was to design a set of electronic dice that would exhibit random behavior. Something that could be made to exhibit pseudo-random behavior (the same sequence could be easily recreated by starting with the proper seed) was not acceptable. I built the dice with a pair of modulo 6 counters that were driven by a high speed clock (typical speed of 500KHz). The high speed clock's frequency was modulated by an unrelated and very slow clock, with a period of several seconds. A person could press the "roll" button and hold it indefinately, though upon being released the "roll" would continue for about 1/2 a second. This roll was assumed to be a fixed length, though it wasn't required to be. The end result was a system that appeared very random. This randomness was based on the inability of a human to press the switch at exactly random intervals though, a machine closing the switch in a very exacting way would produce a pseudorandom string of rolls based on the product of the machine's rate and the two other unrelated rates in the dice circuit. > Topher Cooper > > USENET: ...{allegra,decvax,ihnp4,ucbvax}!decwrl!pbsvax.dec.com!cooper > INTERNET: cooper%pbsvax.DEC@decwrl.dec.com > > Disclaimer: This contains my own opinions, and I am solely responsible for > them. -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Dave Haynie {caip,ihnp4,allegra,seismo}!cbmvax!daveh "Laws to supress tend to strengthen what they would prohibit. This is the fine point on which all the legal professions of history have based their job security." -Bene Gesserit Coda These opinions are my own, though for a small fee they may be yours too. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cipher@mmm.UUCP (Andre Guirard) (11/11/86)
In article <3782@columbia.UUCP> zdenek@heathcliff.columbia.edu.UUCP (Zdenek Radouch) writes: >>From: "Keith F. Lynch" <KFL%MX.LCS.MIT.EDU@MC.LCS.MIT.EDU> >> >> A true random number generator might be a speck of radium with a >>Geiger counter next to it. A silicon device, i.e. computer, can only >>be a pseudo-random number generator. The numbers that are produced by >>it may be well distributed, but they only APPEAR random - because of >>our lack of knowledge of the algorithm.... > >Sorry, the life isn't that simple. Let's not argue about true or pseudo >random numbers, philosophers will do it for us. No, let's argue about true and pseudo random numbers. The distinction really is simpler than you apparently think. If your RNG is deterministic, i.e. if by knowing its current state you can deduce its future state, then it's a pseudo-RNG. A digital computer running a random number program, assuming the hardware is working correctly, falls into this category. A true RNG is by definition unpredictable. An RNG based on phenomena such as atomic decay, which most physicists _believe_ to be non-deterministic, falls into this category. The important distinction for a PSI experiment is that if you use a pseudo-RNG, you can't absolutely rule out the possibility that the subject knows or has (perhaps subconsciously) deduced the algorithm which generates the numbers. And the other important point, which you cut out of your quoting from the other article, is that if a subject can influence the outcome of a _deterministic_ process then there's some funny business going on. A person who can do this should also be able to get the following program: do forever { print "1" } to print something other than "1". Or an easier task, to get a computer to crash just by concentrating at it. This is easy to do by just randomly changing memory locations, after all. >We use computers because it's easy to get any distribution we want and >the process is controlled. If you don't understand the mechanism of >generating the numbers you just get some numbers and you have to analyse >them. And this is fine for statistics experiments and simulations. Not so good, as I said, for PSI experiments where the true source of the "randomness" is very important. > As a result you could say they have such and such distribution, but >that doesn't imply that the numbers to be generated next will have the >same property. But we hope that they will!!!-) >As far as the examples you used, a Geiger counter counts 0,1,2,3,4,5,6.... >That sequence can hardly be considered to be random. The only way to make >it what you call "true random number generator" would be to build some >fairly sophisticated hardware around it. Dum de dum... It was fairly clear to me that this is what was meant. > Even then it wouldn't generate >the numbers with required distribution; it's simply too complicated. I can think of a fairly easy way to generate numbers with a uniform distribution this way. Then you use those as input to an algorithm (see here's where the computer should come in) that uses them to generate the distribution you want. >... A mathematical model is just the easiest and most reliable >approach. True. Unfortunately in PSI experiments there are factors which would cause us to reject the easiest and most reliable way for something more theoretically sound. After all, much PSI research is, in a sense, research into the nature of probability. For this we want to use "real" probability and not a computer simulation. "Oh come on! You don't really need a cyclotron for your particle physics experiments. You can simulate all that stuff on a computer!" ===+=== Andre Guirard /@ @\ ihnp4!mmm!cipher /_____\ ( @ @ ) "Well, yes, I'm opinionated. But only about \ _ / those subjects where I know I'm right." `-'
zdenek@heathcliff.columbia.edu (Zdenek Radouch) (11/11/86)
In article <979@cbmvax.cbmvax.commodore.COM> daveh@cbmvax.commodore.COM (Dave Haynie) writes: > >I used something related to this for a design project in an EE class... > ....I built the dice with >a pair of modulo 6 counters that were driven by a high speed clock (typical >speed of 500KHz). The high speed clock's frequency was modulated by an >unrelated and very slow clock, with a period of several seconds.... > ....This randomness was based >on the inability of a human to press the switch at exactly random intervals... Just a quick question. Was the slow clock modulation result of some analysis or just a means to make the "random" numbers more random? zdenek ------------------------------------------------------------------------- Men are four: He who knows and knows that he knows, he is wise - follow him; He who knows and knows not that he knows, he is asleep - wake him; He who knows not and knows that he knows not, he is simple - teach him; He who knows not and knows not that he knows not, he is a fool - shun him! zdenek@CS.COLUMBIA.EDU or ...!seismo!columbia!cs!zdenek Zdenek Radouch, 457 Computer Science, Columbia University, 500 West 120th St., New York, NY 10027
daveh@cbmvax.cbm.UUCP (Dave Haynie) (11/12/86)
> > In article <979@cbmvax.cbmvax.commodore.COM> daveh@cbmvax.commodore.COM (Dave Haynie) writes: >> >>I used something related to this for a design project in an EE class... >> ....I built the dice with >>a pair of modulo 6 counters that were driven by a high speed clock (typical >>speed of 500KHz). The high speed clock's frequency was modulated by an >>unrelated and very slow clock, with a period of several seconds.... >> ....This randomness was based >>on the inability of a human to press the switch at exactly random intervals... > > Just a quick question. Was the slow clock modulation result of some > analysis or just a means to make the "random" numbers more random? > > zdenek The slow clock itself was the best way I could find to introduce randomness. The components allowed in the circuit were very limited, and the main point of the exercise was to come up with some reasonable way to do the roll randomly. The time periods were chosen with some thought. The fast clock was picked so that even small changes in the frequency would cause a large change in the number of counts through the two die counters. The fixed "roll" time was set at around a 1/2 second mainly because that time looked good, it appeared to most people to be a "good roll"; a shorter time would have been OK, but not as ergonomic. The slow clock time was chosen to be quite around 10 time the length of the fixed roll, but no direct multiple of this roll. This scheme turned out to work very well, though I never did any formal analysis of its "randomness. -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Dave Haynie {caip,ihnp4,allegra,seismo}!cbmvax!daveh "Laws to supress tend to strengthen what they would prohibit. This is the fine point on which all the legal professions of history have based their job security." -Bene Gesserit Coda These opinions are my own, though for a small fee they may be yours too. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
zdenek@heathcliff.columbia.edu (Zdenek Radouch) (11/14/86)
In article <1090@mmm.UUCP> cipher@mmm.UUCP (Andre Guirard) writes: > >...And the other important point, which you cut out of your quoting from >the other article, is that if a subject can influence the outcome of a >_deterministic_ process then there's some funny business going on. A >person who can do this should also be able to get the following program: > > do forever { > print "1" > } > >to print something other than "1". Or an easier task, to get a >computer to crash just by concentrating at it. This is easy to do by >just randomly changing memory locations, after all. [the rest of the article full of PSI deleted] Let me tell you something. I am not an open minded person. I really do have problems with my digesting tract when I see the words "PSI" and "research" right next to each other. This time it's not a joke. Sorry to disappoint you. zdenek zdenek@cs.columbia.edu or ...!seismo!columbia!cs!zdenek