About a month ago Martz, while in a thread with Alex Williams=20
posted this:
As far as I can see we can recursively break all communication down to
these three steps;
1. Something I want to express is converted [1] into a set of symbols
that I think will convey my meaning.
2. These symbols are placed onto some physical medium where the intended
recipient(s) can access them.
3. The symbols are converted [2] by you into something meaningful.
With step 1, the most important step [3], a lot can go wrong. Some
potential problems relate to the media to be used. For example, informal
languages are very imprecise. Ambiguosity abounds in the english
language so to avoid errors we must try to avoid these ambiguosities, an
impossible task. We have designed formal languages which can express a
concept in a singularly precise way (I realise that we interpret even
these within our own memetic ecologies but I consider that a separate
issue, more on which later). However these languages tend to be very
specialised; maths, engineering etc.; none are capable of expressing the
full range of human communication [4]. Another problem is that it's
difficult to keep track of all the media you're actually using to
communicate. Words, body language, tone, pitch etc. all combine to
transmit a high-bandwidth signal. On a minute-to-minute basis, most
people aren't even aware of the signals they're sending, never mind
being practised in the skill of controlling them. I think a lot of this
can be overcome with study and discipline but not everyone will put that
much time and effort into their communications. There comes a point when
the cost of error prevention outweigh the value - how much data
redundancy can you afford to build in? Finally, we build our
communications using our internal map of the universe and as such we
likely each have a totally unique language. All we can do here is to try
to tailor our communication to the internal map of the intended
recipient. No mean feat; we have only our internal map of their internal
map to work from; what if they're also tailoring their interpretation to
take account of their internal map of your internal map (Catch 22); what
to do with an audience of more than one person [5]?
With step 2, the problems are media-dependant. Each medium has its
noise-to-signal ratio. These are beyond my current scope but I'd be
curious to hear any thoughts on what those ratios might be for signals
directly processed by the human senses.
Step 3 has a similar set of problems to step 1.
Will that do to start us off?
[1] I've used the word 'converted' very deliberately here as recent
thoughts (see my question re: wildfire mutation rates in a closed
memetic environment on another thread) have led me to believe that there
is some sort of a feedback loop involved in this process such that just
as our mental structures create the symbols of expression (imperfectly),
so do the symbols we use affect the structures which created them. I'm
still fermenting this one so all ideas gratefully received.
[2] Again, I think I see a feedback loop here. Our interpretation
overlays itself to some extent on the symbols we received, altering them
to better fit the interpretation.
[3] Small errors at source can be magnified the further they travel in a
gemotric fashion. A flaw of +/- 1 degree will amount to +/- 18cm at a
distance of 1m (hasty calculation, correct me if necessary), I think the
same principle applies to communication. We can't map this yet because
we don't even have a measure of 'distance' which can be applied. This
would depend on media the signal was travelling through.
[4] It could be argued that the full range *requires* that the
imperfections be present, but it would be nice to have the choice of
whether to use formal or informal symbols as appropriate.
[5] We would have to tailor our symbols to match only those parts of the
audiences internal maps which they had in common. This would explain why
blockbuster movies tend to have that LCD appeal.
Then Tony posted this:
In message <HT9xUEABnXGzEwUU@martz.demon.co.uk>, Martz
<martz@martz.demon.co.uk> writes
>
>[1] I've used the word 'converted' very deliberately here as recent
>thoughts (see my question re: wildfire mutation rates in a closed
>memetic environment on another thread) have led me to believe that there
>is some sort of a feedback loop involved in this process such that just
>as our mental structures create the symbols of expression (imperfectly),
>so do the symbols we use affect the structures which created them. I'm
>still fermenting this one so all ideas gratefully received.
Dennett talks about this doesnt he? In fact it is one of the
foundations of his theory of how consiousness evolved (and continues to
do so.). I wish I'd seen your Wildfire mutation rates thread. =20
>
>[2] Again, I think I see a feedback loop here. Our interpretation
>overlays itself to some extent on the symbols we received, altering them
>to better fit the interpretation.
This is exactly the same as [1] except in reverse.
>
>[3] Small errors at source can be magnified the further they travel in a
>gemotric fashion. A flaw of +/- 1 degree will amount to +/- 18cm at a
>distance of 1m (hasty calculation, correct me if necessary), I think the
>same principle applies to communication. We can't map this yet because
>we don't even have a measure of 'distance' which can be applied. This
>would depend on media the signal was travelling through.
Yes I see. Heres some distances.
1)Photocopier: Zero distance
2)spoken conversation between A & B: Big distance
3)Written conversation between A & B: smaller distance
in 2) and 3) the effective distance decreases if the exchanges are
itterated.
The feedback process you talked about in the conversions helps
reduce the distance as well.
In written conversations two linguistic scholars reduce
the distance between them by having perfect syntax (or is the correct
word protocol?). Idealy there Syntactic distance can be zero. Semantic
distance only ever reaches zero on what we call "perfect agreement"=20
Perfect logic is a zero distance comunication tool.
Hey it would be great fun to try and agree on a distance scale.
=20
=20
>
>[4] It could be argued that the full range *requires* that the
>imperfections be present, but it would be nice to have the choice of
>whether to use formal or informal symbols as appropriate.
I think imperfections can be a curse and a blessing. The
parallell with biological mutations is strong here.
>
>[5] We would have to tailor our symbols to match only those parts of the
>audiences internal maps which they had in common. This would explain why
>blockbuster movies tend to have that LCD appeal.
Ok I have spent ages error correcting and now anything further
is diminishiing returns, especially since I have only this one posting
from you with which to biuld my mental model of you, with which to
correct my errors here.=20
I hope you dont mind this intrusion but I would love to reduce
our semantic distance. This must be an important objective.
In message <TlLuNBAmdQHzEwsD@joney.demon.co.uk>, Tony Hindle=20
<t.hindle@joney.demon.co.uk> writes
>In message <ZQdGUUAD90GzEwmj@martz.demon.co.uk>, Martz=20
><martz@martz.demon.co.uk> writes
>>On Mon, 3 Mar 1997, Tony Hindle <t.hindle@joney.demon.co.uk> wrote:
>>
>>>>[1] I've used the word 'converted' very deliberately here as recent
>>>>thoughts (see my question re: wildfire mutation rates in a closed
>>>>memetic environment on another thread) have led me to believe that there
>>>>is some sort of a feedback loop involved in this process such that just
>>>>as our mental structures create the symbols of expression (imperfectly),
>>>>so do the symbols we use affect the structures which created them. I'm
>>>>still fermenting this one so all ideas gratefully received.
>>> Dennett talks about this doesnt he?=20
>>
>>Where? I'm woefully behind on my reading list at the moment, to the
>>point where I've had to start culling. A brief summary of his ideas?
>
>
> I think it=92s Darwin=92s Dangerous idea (if not it=92s=20
>consciousness explained).
> I havent got a copy at hand so this is as I remember it.
>Dennett talks about words (symbols of expression) being=20
>(imperfect)=93mind tools=94. we use them internaly as we think (speak=20
>to ourselves). A spoken/written sentence is really the end of a=20
>translation process from the original thought. (the current mind=20
>tool set of the brain does the translation). This translation is=20
>an aproximation to the meaning of the original thought (a clue).=20
>The translation then acts on the brain as a powerfull reminder of=20
>the original thought which is then translated again to produce=20
>another =93example translation=94 this operation cycles a few times=20
>until there are many example translations that aproximate the=20
>original thought. There will be an abstract class to which they=20
>will all belong and this will be the original thought, which the=20
>brain can label. The label is now available as a new word or a new=20
>mind tool for use in the future.
>
> Ok, Im going to try again; The bit that your =93feedback=20
>loop=94 reminded me of is when he describes how consciousness=20
>bootstraped itself into existence. He talks about language being=20
>a mental prosthetic (as a wooden leg is a physical one). So when a=20
>new idea occurs to you, you initialy verbalise/write it to=20
>yourself. Another way of saying this is you =93think that thought=20
>with any words you already possess. (call this draught 1, Im=20
>starting in the middle I know) . Now you =93say/read it back to=20
>yourself=94 and what it actually means to you is not exactly what=20
>your initial thought was (even though your initial thought is now=20
>changed slightly as you =93hear what you have said=94) so you try to=20
>correct/improve it and produce draught 2. Now your thinking acts=20
>upon draught 2 and =93sees=94 it is not exactly what you meant and so=20
>on. Is this the feedback process you were describing.
> Eventualy you have a (long) description that conveys the=20
>meaning of your initial thought, call this description X. You have=20
>now created a new word X. Your mental tool set has been improved,=20
>next time you have a new idea you have a finer grained mental tool=20
>kit with which to describe it before giving it a name and thus=20
>creating your next word. =20
> =20
> Well that's three atempts Ive had to explain it, maybe you=20
>can abstract something from them.
>>
>>>In fact it is one of the
>>>foundations of his theory of how consiousness evolved (and continues to
>>>do so.). I wish I'd seen your Wildfire mutation rates thread. =20
>>
>>In short; take three women [1], tell one of them something, put them in
>>a room together for a few hours (preferably spread over a longer period,
>>a couple of days perhaps) and whatever you told them has mutated to the
>>point where *none* of them can remember the original, including the
>>person you told it to. It occurred to me afterwards that reminding them
>>can work sometimes but can just as easily get you lynched.
>>
>>>>[3] Small errors at source can be magnified the further they travel in a
>>>>gemotric fashion. A flaw of +/- 1 degree will amount to +/- 18cm at a
>>>>distance of 1m (hasty calculation, correct me if necessary), I think the
>>>>same principle applies to communication. We can't map this yet because
>>>>we don't even have a measure of 'distance' which can be applied. This
>>>>would depend on media the signal was travelling through.
>>> Yes I see. Heres some distances.
>>> 1)Photocopier: Zero distance
>>
>>Not true. Try a 40th generation photocopy and you'll see the difference.
> Point taken. A perfect photocopier would have zero=20
>distance. This reminds me of digital signal transmission and=20
>repeater stations. If an analogue signal is used for transmission=20
>extra noise will be added while the signal is in the transmission=20
>line. For a digital signal provided the "noise" on the transmission=20
>line doesnot exceed a threshold level it can all be removed by=20
>decoding/recoding and amplifying at repeater stations. that is why=20
>they have repeater stations every so often in the transmission=20
>lines. Analogous to this would be a photocopier being used for say=20
>6 generations, then a human scrupulously checking letter, by letter=20
>the print on the sixth generation copy and going over all the=20
>letters in black ink. In this context I suppose a zero distance=20
>device would be a text copier, at least it would be zero distance=20
>for the text if it had perfect protocol for that text (easily done=20
>digitally).=20
>>
>>> 2)spoken conversation between A & B: Big distance
>>
>>And higher bandwidth. More signal. Even more if they're in eyesight.
> Yes. Yes. yes (although under some circumstances eyesight=20
>can increase the distance.) You see a beautiful woman you want to=20
>comunicate something impressive which you rehearse. Then as face=20
>one another, you say "Agh-blughl-oom".
>>
>>> 3)Written conversation between A & B: smaller distance
>>>in 2) and 3) the effective distance decreases if the exchanges are
>>>itterated.
>>
>>Agreed up to a point.
> That'll do for now.
>
>>
>>> The feedback process you talked about in the conversions helps
>>>reduce the distance as well.
>>
>>I disagree strongly with this. I think it *increases* the distance. It's
>>like mapping a big chinese whispers chain onto two people. There's
>>better error correction but it's still not perfect and every time the
>>idea travels the loop it has greater potential for distortion.
> I think we have to distinguish two cases. Lets take a line=20
>of chinese whispers as an example. As the meme went from left to=20
>right it would be increasing distance, but if each exchange was=20
>allowed a few cycles of feedback the effective distance would be=20
>decreased. Indeed if the meme was a single word and lots of=20
>feedback was allowed the line could have people of all different=20
>nationalities and a single word meme could get through, especially=20
>if every sixth person was a repeater which in this context would be=20
>someone who was a native speaker of the meme.
> A second case would be when the meme was a long statement,=20
>say a one minute monologue rich with meaning. I agree that feedback=20
>at each stage in this case would cause extra distance from the=20
>original meme but i dont think the distance would keep on=20
>increasing (see below.)
> .SOMETHING HAS JUST OCCURRED TO ME ON RE-READING THIS. THE=20
>DISTANCE DIMENSION WE ARE DISCUSSING IS REALLY A DISTANCE IN N=20
>DIMENSIONAL THOUGHT SPACE OR BABELS LIBRARY AS DENNETT CALLES IT.
> As for mapping the chinese whispers onto two people, Yes I=20
>can see that The original meme could see more distance with more=20
>iterations but I think it would be moving towards "a basin of=20
>attraction". At the basin would be a more stable (=3Dsatisfying?)=20
>meme.
> For the three women talking I think it would end up as a=20
>mutualy agreed tale that made all three of them feel good. If it=20
>also compelled them to tell others it would be something we might=20
>expect to hear often (gossip-with the spin that suits them).
> With us during our exchenges the meme will also move=20
>towards a basin of attraction, more satisfying to both of us=20
>because it complements our conceptual tool kit......Women eh!
>
>
>
>>
>>>Semantic
>>>distance only ever reaches zero on what we call "perfect agreement"=20
>>
>>Even then, I'm agreeing to what I *think* you're saying and vice-versa.
>>I agree with Alex here in that I don't think it *ever* reaches zero.
> Point taken. When we agree we both agree with what we think=20
>the other is saying. The beauty of further comunication is that we=20
>can check our agreement by experiments e.g., use the new meme in a=20
>new context and see if the other understands.
>
>>> Perfect logic is a zero distance comunication tool.
>>Hmmm. I don't think so, but you're welcome to try to convince me.
> I supose I meant perfect protocol with no transmission=20
>noise. =20
>>
>>>>[4] It could be argued that the full range *requires* that the
>>>>imperfections be present, but it would be nice to have the choice of
>>>>whether to use formal or informal symbols as appropriate.
>>> I think imperfections can be a curse and a blessing. The
>>>parallell with biological mutations is strong here.
>>
>>Very. That's why I'm totally against dropping the biological metaphor,
>>when it holds, it holds good. Alex has a point in that we mustn't become
>>constrained by it though.
> Any metaphor that is partly isomorphic (meaning as I=20
>remember Hofstadter's meaning) is usefull but we must know its=20
>limits (exceeding them can be good for a laugh though)
>>
>>>>[5] We would have to tailor our symbols to match only those parts of the
>>>>audiences internal maps which they had in common. This would explain why
>>>>blockbuster movies tend to have that LCD appeal.
>>>
>>> Ok I have spent ages error correcting and now anything further
>>>is diminishiing returns
>>
>>Awwwww. You were just getting to the good bit.
> was there more, send me it please, before I go away. I must=20
>have deleted part of the original it or something.=20
>>
In message <KGiihHA6qyHzEwsU@martz.demon.co.uk>, Martz=20
<martz@martz.demon.co.uk> writes
>On Wed, 5 Mar 1997, Tony Hindle <t.hindle@joney.demon.co.uk> wrote:
>
><snipped Dennet summary>
>
>Yep. That's pretty much what I meant. Buggerit.
>
><snipped internal feedback loop>
>
>I can agree that when it's internal the attractor that is your memory of
>something may error-correct.
>
>>In this context I suppose a zero distance=20
>>device would be a text copier, at least it would be zero distance=20
>>for the text if it had perfect protocol for that text (easily done=20
>>digitally).=20
>
>It is??? Do tell.
>
>>>> 2)spoken conversation between A & B: Big distance
>>>
>>>And higher bandwidth. More signal. Even more if they're in eyesight.
>> Yes. Yes. yes (although under some circumstances eyesight=20
>>can increase the distance.) You see a beautiful woman you want to=20
>>comunicate something impressive which you rehearse. Then as face=20
>>one another, you say "Agh-blughl-oom".
>
>Heh. Good point.
>
>>>> The feedback process you talked about in the conversions helps
>>>>reduce the distance as well.
>
><snipped error reduction using repeaters>
>
>Alex would say that the meme hasn't been transmitted at all, there's
>still the translation from ear to brain to go but I take your broad
>point. A repeater could reduce the number of errors.
>
>> .SOMETHING HAS JUST OCCURRED TO ME ON RE-READING THIS. THE=20
>>DISTANCE DIMENSION WE ARE DISCUSSING IS REALLY A DISTANCE IN N=20
>>DIMENSIONAL THOUGHT SPACE OR BABELS LIBRARY AS DENNETT CALLES IT.
>
>OK. That's put Dennett a couple of steps closer to the READ THIS NOW
>spot.
>
>> As for mapping the chinese whispers onto two people, Yes I=20
>>can see that The original meme could see more distance with more=20
>>iterations but I think it would be moving towards "a basin of=20
>>attraction". At the basin would be a more stable (=3Dsatisfying?)=20
>>meme.
>
>This makes sense. It just amazes me the distance it can cover before
>settling. Unrecognisable at times.
>
>> For the three women talking I think it would end up as a=20
>>mutualy agreed tale that made all three of them feel good. If it=20
>>also compelled them to tell others it would be something we might=20
>>expect to hear often (gossip-with the spin that suits them).
>> With us during our exchenges the meme will also move=20
>>towards a basin of attraction, more satisfying to both of us=20
>>because it complements our conceptual tool kit......Women eh!
>
>You've said a mouthful there. I'd shied away from putting 'gossip' and
>'women' in the same paragraph. You're braver than I.
>
>> Point taken. When we agree we both agree with what we think=20
>>the other is saying. The beauty of further comunication is that we=20
>>can check our agreement by experiments e.g., use the new meme in a=20
>>new context and see if the other understands.
>
>But we can never prove perfect understanding to be true, we can only
>prove it false. So even if we get there we won't know for sure.
>
>>>> Perfect logic is a zero distance comunication tool.
>>>Hmmm. I don't think so, but you're welcome to try to convince me.
>> I supose I meant perfect protocol with no transmission=20
>>noise. =20
>
>It's a theoritical construct which I don't think we'll ever find a match
>for in nature (to my mind it's also verging on tautology but perhaps
>that's just perspective) but yes, it is.
>
>> Any metaphor that is partly isomorphic (meaning as I=20
>>remember Hofstadter's meaning) is usefull but we must know its=20
>>limits (exceeding them can be good for a laugh though)
>
>Extrapolatio ad absurdum?
>
>>>Awwwww. You were just getting to the good bit.
>
>I meant depth-wise; spend more time error-correcting...the good bit's in
>there if you can find it. Just a bit of banter.
>
>>> =20
I hope you find some usefull memes in all this for your analysis=20
Corey. Sorry it's all mixed up a bit.
Tony Hindle.
Telling the world about Martz's 3 point recursive comunication model