Freethought & Rationalism Archive

Francois Tremblay · 02-05-2002, 10:26 AM

Mageth, these patterns can be interpreted meaningfully. That is why we say they are information. Understanding, for example, the meaning of tree rings requires knowledge of the biology of trees, and so on. Without this knowledge, the pattern of tree rings is not actual information (note I said actual).

As I said, information is both dependant on the substrate itself (the patterns that we label as information) and the interpreting mechanism (that gives it meaning). You cannot have information without these two characteristics.

A more rigorous definition of "information" in science is the amount of possible choices that it rules out - reduction of uncertainty (the classical example of saying that a person is a man or a woman is the resolution of 1 binary choice). And you can see that to determine the choices that are resolved by a piece of information, you need to understand the decoding mechanism in the first place.

Suppose we want to analyze the amount of information contained in a MIDI file (which is a synthetic music format). What kind of parameters do we need to determine how much information is in the file ?

* the length and complexity of the file (most likely thru a decoding program) - i.e. the pattern itself.
* the number of possibilities for notes and number of instruments for any MIDI file (which is an attribute of the MIDI file format, and by extension, of the decoder) - i.e. the interpreting mechanism.

Thus, by knowing how many possibilities exist, and how many possibilities are indeed fulfilled by the file, we end up with an assessment of the degree of information contained in the file. If the MIDI file format is as unredundant as possible, then the number of bits of the file would be a good indication of this measure.

[ February 05, 2002: Message edited by: Franc28 ]

HRG · 02-05-2002, 10:57 AM

Quote:

Originally posted by David Gould:


I understand your point about keeping the meaning consistent.

However, I disagree that information can be destroyed. Shredding it merely makes it more difficult to access - the paper can be put back together with the input of some time and energy and thus the message is not lost.

I agree that there are examples (burning of the paper, for example) that would make it extremely difficult to reconstruct the message.

However, I do not see that it is impossible in principle as all physical processes are reversible.

While general relativity is not necessarily the last word on black holes, let me point out that the transit into the event horizon of a black hole is irreversible. Black holes are the ultimate information shredders, at least in classical relativity.

There was a debate among gravitationists whether this information is partially recovered via Hawking radiation, IIRC.

Regards,
HRG.

Happy Wonderer · 02-05-2002, 09:19 PM

Quote:

Originally posted by David Gould:

In your shredder example, it is certainly possible to put the pieces back together. Thus, the information has not been destroyed.

As all known physical processes are reversible, it is hard to see how information as such can be destroyed - made harder to find, for sure, but not destroyed.

David

I was thinking about this today and wondering what a good definition of information destruction would be. My shredder example was originally intended to be humorously topical, but I think it does have some potential to lead to a definition.

The most famous case of reconstructing shredded documents was when the Iranian students took over the US Consulate in Iran. They were able to reconstruct documents (shredded in haste using inexcusably bad technology) by having skilled carpet weavers arrange the strips of documents so that they made sense. (The shredder the consulate used shredded the documents only vertically, leaving strips of letters.)

In this case, they could be fairly confident that they had reconstructed the documents correctly, because the strips would line up correctly based upon paragraph spacing. They could tell that they had complete words and sentences that made sense. Their confidence level could not be %100, though, it is possible that two documents (especially short ones or tables of numbers) could have a false match, and the words happened to make sense. They were also probably helped by the fact that many of the layers of paper would be in the correct order, since the documents were hastily shredded in bulk.

High-security shredders have an extra set of knives that dice the strips into squares the size of a single letter or smaller. You are now left with something that looks more like kitty litter than confetti. The idea is that you lose both horizontal and vertical positioning of the letter. Even if you can figure out which sheet of paper a set of letters belongs to, there is a very large number of potential configuations, many of which make sense. This is crucial, the older technology has a smaller number of configurations per sheet and likely only one is going to make sense. Even if you do incredibly detailed microscopic evaluation, the odds of making a mistake are extremely high; especially considering that as part of the process of reconstruction you are going to have to interpret the data, which will invariably lead to contamination.

What does this digression have to do with the price of tea in Iran? We may be coming upon a measure of information loss (destruction.) Information is "more destroyed" the less confident that any viewer is that they have something that matches the original signal. As this confidence level approaches (or reaches) zero, I'd argue that the information was destroyed. This view of information may be different than that of a physicist.

You could possibly turn this definition around to define information creation. I'd agree with you that ideas are mostly based upon existing information, but I'd not agree that they themselves are pre-existing. A way to define new information may be to describe the confidence that you would be able to predict the new information based upon existing. The lower that confidence, the more likely that this is truly novel information.

We have a slight difference in approach to my example. In my username/password case, I consider the new information to be "the XYZZY connection function specifies user name before password -- All other possibilities are not part of the XYZZY spec." The fact that the spec was done that way may be completely unpredictable and related to who happened to be in the design meeting at a crucial time. (Once decisions are made they take on a life of their own.) In this case, since the outcome is unpredictable (except to a complete determinist), I'd argue that the statement above is new information.

In other threads the discussion seems to be "is it information if there isn't a decoder?" I certainly think 'it' is information if there exists (existed?) the possibility of a decoder. We are now able to take core samples from very deep in the earth to figure out what was going on (at some degree of confidence) before any life existed. The information about an asteroid hit has been there all along, whether or not humans eventually came around to interpret it.

However, if there is no possibility of a decoder (how?) is it information? Not a very useful question but an interesting one...

You make the statement that "all known physical processes are reversable." I'm not a hard science person, so I'd like to know more about that. It doesn't comport to my layman's experience (but neither does any of modern physics.) It seems like there is always going to be uncertanty, friction, the third law, and other stuff to mess you up when trying to reverse a process.

This is an interesting thread. It isn't going where I thought it would. The science people have it easy -- everybody knows what "Energy" is. I think it is much harder to figure out what "Information" is. Just wait until we discuss "Intelligence" (of the human, artificial, or Divine :-) kind.)

Good question!

HW

[ February 05, 2002: Message edited by: Happy Wonderer ]

excreationist · 02-05-2002, 09:53 PM

David Gould:

Quote:

In your shredder example, it is certainly possible to put the pieces back together. Thus, the information has not been destroyed.

What if the documents and everyone whoever had read them were put next to an atomic bomb and it was detonated...? Would the information be able to be fully retrieved?

Quote:

As all known physical processes are reversible, it is hard to see how information as such can be destroyed - made harder to find, for sure, but not destroyed.

Say there was a person that was cut off from the world after they had learnt to read. Let's say that everyday they wrote many stories and then burnt them and ate the stories up (with some food). Their faeces was used to grow plants so that the molecules could be reintegrated into the person's body. The ashes were also used as fertilizer (if that's a good fertilizer).
In the end they wrote thousands of stories. Would it be possible to retrieve all of that information years later, including the exact position of every single pen-stroke?
That might involve simultaneously detecting the molecular state of the room and also the surrounding universe and running a universe simulator program in reverse. (You'd need to simulate the entire universe since the room with the person in it isn't a closed system - there would be gravitational disturbances and quantum entanglement might cause problems)
But to detect the entire state of the physical universe would require a storage capacity of about the same size, and since the computer is part of the universe, it would also have to be included in the simulation. And there would be problems with privacy since some people wouldn't want you to know about their part of the universe, including its history.

Francois Tremblay · 02-05-2002, 10:12 PM

Any type of information can be dissembled in some way and reassembled. It depends on the substrate. There is a point to which we cannot reconstitute the information, but that's more a question of technology than it is of philosophy. For example, we can try to restore old works of art, but the extent to which this restoration will be accurate depends on what we can recuperate from the substrate and our capacity to replicate it.

For example, if you take the case of the MIDI file, deleting that file would, for an average user, make that information destroyed. To an expert in security, that information may be retrievable. So that limit is relative also.

crocodile deathroll · 02-06-2002, 06:25 PM

I have another thought, If a war correspondent had written a document in Hiroshima on 5th of August 1945 and that same document was simultaneously and accidently plagiarized word for word by a war correspondent in Tokyo. Just the mere fact that that document was destroyed by an atom bomb that same day in Hiroshima will make absolutely no difference to the war correspondent in Tokyo . So information is a boundaryless entity it is not at all a unique property of just one place.

crocodile deathroll

excreationist · 02-06-2002, 08:42 PM

crocodile deathroll:
In my atom bomb example, I said that the document and anyone who had read it would be put next to the bomb. And also all copies of the document should be put near the bomb as well.

Francois Tremblay · 02-06-2002, 09:44 PM

Gee crocodile deatroll, let's just add magic, telepathy, teleportation, etc to the equation. You're just making a completely unreal situation.

Since the initial pattern was destroyed anyway, it is much more relevant to treat the new similar pattern as a new one. I am not sure how relevant it is to discuss accidental duplications.

[ February 07, 2002: Message edited by: Franc28 ]

tronvillain · 02-07-2002, 12:20 AM

excreationist, crocodile deathtroll is saying that the document is reproduced exactly independently, but as Franc28 points out, this is ridiculously unlikely.

While information may be "lost" when the medium in which it is realized is destroyed, nothing prevents it from being realized again. This may be extremely unlikely, as in crocodile deathroll's example, or it may not.

[ February 07, 2002: Message edited by: tronvillain ]

excreationist · 02-07-2002, 01:20 AM

Quote:

Originally posted by tronvillain:
excreationist, crocodile deathtroll is saying that the document is reproduced exactly independently, but as Franc28 points out, this is ridiculously unlikely.

Oh... I missed that word in bold somehow...

Ok, so the exact wording of the destroyed document could appear elsewhere by chance... but part of the information involved with the original document is where it was located. Those who wrote the second document can't be sure that there was also a completely identical document written in Hiroshima. I guess they could assert that, and be correct by chance... but people who claim to have knowledge without proof are often mistaken. (Unless they are psychic or something)

02-05-2002, 10:26 AM	#21
Francois Tremblay Banned Join Date: Sep 2000 Location: Montreal, QC Canada Posts: 876	Mageth, these patterns can be interpreted meaningfully. That is why we say they are information. Understanding, for example, the meaning of tree rings requires knowledge of the biology of trees, and so on. Without this knowledge, the pattern of tree rings is not actual information (note I said actual). As I said, information is both dependant on the substrate itself (the patterns that we label as information) and the interpreting mechanism (that gives it meaning). You cannot have information without these two characteristics. A more rigorous definition of "information" in science is the amount of possible choices that it rules out - reduction of uncertainty (the classical example of saying that a person is a man or a woman is the resolution of 1 binary choice). And you can see that to determine the choices that are resolved by a piece of information, you need to understand the decoding mechanism in the first place. Suppose we want to analyze the amount of information contained in a MIDI file (which is a synthetic music format). What kind of parameters do we need to determine how much information is in the file ? * the length and complexity of the file (most likely thru a decoding program) - i.e. the pattern itself. * the number of possibilities for notes and number of instruments for any MIDI file (which is an attribute of the MIDI file format, and by extension, of the decoder) - i.e. the interpreting mechanism. Thus, by knowing how many possibilities exist, and how many possibilities are indeed fulfilled by the file, we end up with an assessment of the degree of information contained in the file. If the MIDI file format is as unredundant as possible, then the number of bits of the file would be a good indication of this measure. [ February 05, 2002: Message edited by: Franc28 ]</p>

02-06-2002, 09:44 PM	#28
Francois Tremblay Banned Join Date: Sep 2000 Location: Montreal, QC Canada Posts: 876	Gee crocodile deatroll, let's just add magic, telepathy, teleportation, etc to the equation. You're just making a completely unreal situation. Since the initial pattern was destroyed anyway, it is much more relevant to treat the new similar pattern as a new one. I am not sure how relevant it is to discuss accidental duplications. [ February 07, 2002: Message edited by: Franc28 ]</p>

02-07-2002, 12:20 AM	#29
tronvillain Veteran Member Join Date: Oct 2000 Location: Alberta, Canada Posts: 5,658	excreationist, crocodile deathtroll is saying that the document is reproduced exactly independently, but as Franc28 points out, this is ridiculously unlikely. While information may be "lost" when the medium in which it is realized is destroyed, nothing prevents it from being realized again. This may be extremely unlikely, as in crocodile deathroll's example, or it may not. [ February 07, 2002: Message edited by: tronvillain ]</p>

Thread Tools	Search this Thread
Show Printable Version	Search this Thread: Advanced Search

02-05-2002, 10:12 PM	#25
Francois Tremblay Banned Join Date: Sep 2000 Location: Montreal, QC Canada Posts: 876	Any type of information can be dissembled in some way and reassembled. It depends on the substrate. There is a point to which we cannot reconstitute the information, but that's more a question of technology than it is of philosophy. For example, we can try to restore old works of art, but the extent to which this restoration will be accurate depends on what we can recuperate from the substrate and our capacity to replicate it. For example, if you take the case of the MIDI file, deleting that file would, for an average user, make that information destroyed. To an expert in security, that information may be retrievable. So that limit is relative also.

02-06-2002, 06:25 PM	#26
crocodile deathroll Veteran Member Join Date: Dec 2001 Location: Darwin Posts: 1,466	I have another thought, If a war correspondent had written a document in Hiroshima on 5th of August 1945 and that same document was simultaneously and accidently plagiarized word for word by a war correspondent in Tokyo. Just the mere fact that that document was destroyed by an atom bomb that same day in Hiroshima will make absolutely no difference to the war correspondent in Tokyo . So information is a boundaryless entity it is not at all a unique property of just one place. crocodile deathroll

02-06-2002, 08:42 PM	#27
excreationist Veteran Member Join Date: Aug 2000 Location: Australia Posts: 4,886	crocodile deathroll: In my atom bomb example, I said that the document and anyone who had read it would be put next to the bomb. And also all copies of the document should be put near the bomb as well.

Freethought & Rationalism Archive

The archives are read only.