Hi all,
A quick question regarding,
pd/doc/5.reference/intro-help.pd
I was wondering why [print~] and [samphold~] are under "AUDIO FILTERS". Are they meant to belong there?
-- David Shimamoto
On Sat, 26 Sep 2009, PSPunch wrote:
Hi all, A quick question regarding, pd/doc/5.reference/intro-help.pd I was wondering why [print~] and [samphold~] are under "AUDIO FILTERS". Are they meant to belong there?
They are meant to remind everybody that categories don't necessarily make any sense.
I'd also ask what's the logic in not putting all the AUDIO FILTERS object classes in the AUDIO MATH section, and/or in not putting all the AUDIO MATH classes in the AUDIO FILTERS section. But I don't expect an answer at all.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
Hi Mathieu,
Hi all, A quick question regarding, pd/doc/5.reference/intro-help.pd I was wondering why [print~] and [samphold~] are under "AUDIO FILTERS". Are they meant to belong there?
They are meant to remind everybody that categories don't necessarily make any sense.
I'd also ask what's the logic in not putting all the AUDIO FILTERS object classes in the AUDIO MATH section, and/or in not putting all the AUDIO MATH classes in the AUDIO FILTERS section. But I don't expect an answer at all.
Haha, cool.
May I take it that there really is no relevancy (as far as you are aware)?
Actually this is question arose while proof reading instruction manuals written by a member of our local user group.
As there is a chance of it being widely circulated, I guess he may have to issue it based on pd-help "as is", and refer to Mathieu's comment if anyone asks the same, although if it was never brought up here, chances of it being asked again may be slim.
-- David Shimamoto
On Sat, 26 Sep 2009, PSPunch wrote:
Hi Mathieu,
I'd also ask what's the logic in not putting all the AUDIO FILTERS object classes in the AUDIO MATH section, and/or in not putting all the AUDIO MATH classes in the AUDIO FILTERS section. But I don't expect an answer at all.
May I take it that there really is no relevancy (as far as you are aware)?
No, I know exactly what the relevancy is, I just don't enjoy it. First, a person tells himself/herself «it would be better if there were categories». Then the person looks for characteristic features of the elements to be categorised, so that categories can be made. Those features have to be easy to think about. Turns out that one of the easiest features to think about in this case, are things like: where you first learned the basic concept of each object class. It's a kind of microcosm of the whole job-title social structure. Let me give an example.
[lop~] is not an operation you learn in elementary-school or high-school math, therefore it doesn't fit in MATH. It doubly doesn't fit in math, because it isn't taught in a Math Department. A Math Department is a social structure that concentrates on any math concept that doesn't belong to any other discipline already, because if Electrical Engineers already occupy the [lop~] land, it's not only redundant for Math Departments to claim it, it also would make Mathematicians look like Electrical Engineers. So not only [lop~] is not part of Math Depts, but a bunch of related topics are just on the border, so they get lumped into a course called Applied Math, which is all made of pure theory, it's just a form of discrimination against kinds of Math that are too much in use by other departments. Meanwhile, Electrical Engineers would say that [lop~] is math, except when they get distracted by a category system. But most of all, for music students, [+~] is true math, whereas [lop~] is something magical and not math, because [lop~] is not part of what they learnt in courses labelled as «math» before, so it looks a lot more «audiosome» than +~ does. This is a summary. The actual situation is more complicated.
So basically the category system has more to do with social factors than with anything else... and those social factors don't help seeing things as they are. For example, something that unites most of AUDIO MATH object classes, is that the effect only involves one instant at a time, no memory, no feedback. This obviously excludes all four [fft~] and [framp~] from that category system, as those are block-oriented object classes (which could be the name of another category). But then, there are a few expatriates that you have to pick from all over to put them in the instant-oriented category. For example, [cos~] from the OSCILLATORS AND TABLES category; but also, the [tabread...] classes are instant-oriented, but they differ from all others so far, because they use data that doesn't come from the signal. Then we could argue about whether [noise~] belongs in or not (because it depends on how you look at it).
I'm not completely against categories... I'm trying very hard to make good categorisations, because it's hard for me to find a categorisation that I can take seriously, and I'm trying to find one.
As there is a chance of it being widely circulated, I guess he may have to issue it based on pd-help "as is", and refer to Mathieu's comment if anyone asks the same,
At this point, I don't expect Pd's category list to change at all, so, depending on what it is that you're doing, it may be better to just go with Pd's categories, if you have any advantage in following Pd's categories.
although if it was never brought up here, chances of it being asked again may be slim.
Oh, the general topic was brought here in the past. For example, I remember some years ago there was a thread about whether [namecanvas] is OBSOLETE or not. It's not. (As you see, it didn't change Pd's official categorisations).
But also, for each post to the pd-list, there may be 10 or 100 people asking themselves the same thing, roughly speaking. You don't know. In any case, downloads of pd-extended aren't on the same scale as the member-list of pd-list, and then, not everybody ever writes at all.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
Hi Mathieu,
Thank you for sharing your thoughts. I fully understand what you find unpleasant with categorization in general.
Perhaps I should have made clear that I was not sure how [print~] and [samphold~] would even be considered to go under filters.
Following your example, I figure [print~] may have been heavily used for tracking outputs while designing filters which lead it to belong under the category.
Is [samphold~] also often used in building filters?
-- David Shimamoto
Hi Mathieu,
I'd also ask what's the logic in not putting all the AUDIO FILTERS object classes in the AUDIO MATH section, and/or in not putting all the AUDIO MATH classes in the AUDIO FILTERS section. But I don't expect an answer at all.
May I take it that there really is no relevancy (as far as you are aware)?
No, I know exactly what the relevancy is, I just don't enjoy it. First, a person tells himself/herself «it would be better if there were categories». Then the person looks for characteristic features of the elements to be categorised, so that categories can be made. Those features have to be easy to think about. Turns out that one of the easiest features to think about in this case, are things like: where you first learned the basic concept of each object class. It's a kind of microcosm of the whole job-title social structure. Let me give an example.
[lop~] is not an operation you learn in elementary-school or high-school math, therefore it doesn't fit in MATH. It doubly doesn't fit in math, because it isn't taught in a Math Department. A Math Department is a social structure that concentrates on any math concept that doesn't belong to any other discipline already, because if Electrical Engineers already occupy the [lop~] land, it's not only redundant for Math Departments to claim it, it also would make Mathematicians look like Electrical Engineers. So not only [lop~] is not part of Math Depts, but a bunch of related topics are just on the border, so they get lumped into a course called Applied Math, which is all made of pure theory, it's just a form of discrimination against kinds of Math that are too much in use by other departments. Meanwhile, Electrical Engineers would say that [lop~] is math, except when they get distracted by a category system. But most of all, for music students, [+~] is true math, whereas [lop~] is something magical and not math, because [lop~] is not part of what they learnt in courses labelled as «math» before, so it looks a lot more «audiosome» than +~ does. This is a summary. The actual situation is more complicated.
So basically the category system has more to do with social factors than with anything else... and those social factors don't help seeing things as they are. For example, something that unites most of AUDIO MATH object classes, is that the effect only involves one instant at a time, no memory, no feedback. This obviously excludes all four [fft~] and [framp~] from that category system, as those are block-oriented object classes (which could be the name of another category). But then, there are a few expatriates that you have to pick from all over to put them in the instant-oriented category. For example, [cos~] from the OSCILLATORS AND TABLES category; but also, the [tabread...] classes are instant-oriented, but they differ from all others so far, because they use data that doesn't come from the signal. Then we could argue about whether [noise~] belongs in or not (because it depends on how you look at it).
I'm not completely against categories... I'm trying very hard to make good categorisations, because it's hard for me to find a categorisation that I can take seriously, and I'm trying to find one.
As there is a chance of it being widely circulated, I guess he may have to issue it based on pd-help "as is", and refer to Mathieu's comment if anyone asks the same,
At this point, I don't expect Pd's category list to change at all, so, depending on what it is that you're doing, it may be better to just go with Pd's categories, if you have any advantage in following Pd's categories.
although if it was never brought up here, chances of it being asked again may be slim.
Oh, the general topic was brought here in the past. For example, I remember some years ago there was a thread about whether [namecanvas] is OBSOLETE or not. It's not. (As you see, it didn't change Pd's official categorisations).
But also, for each post to the pd-list, there may be 10 or 100 people asking themselves the same thing, roughly speaking. You don't know. In any case, downloads of pd-extended aren't on the same scale as the member-list of pd-list, and then, not everybody ever writes at all.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
On Sat, 26 Sep 2009, PSPunch wrote:
Perhaps I should have made clear that I was not sure how [print~] and [samphold~] would even be considered to go under filters.
No, no, you were clear enough, I just meant that those aren't the only things that don't fit where they've been put, and that the problem is much bigger than that. (e.g. afaik, in Signal Theory, [rpole~] is not a "filter", though it still is peripherally related to filters; otoh there might be other Signal Theorists using different definitions or namings).
Is [samphold~] also often used in building filters?
I don't know... but it isn't "filtering" because what you can get out of it can have a richer spectrum than the original (left-inlet input), and it isn't "linear" either, or "quad", or whatever... it doesn't fit the filter theory much... and I don't see how using it anywhere inside an abstraction can not prevent the abstraction to be a linear filter or quad filter...!
According to my num.analysis book, [samphold~] would be called a "piecewise-constant interpolator", with the warning that "constant interpolator" is somewhat a contradiction of terms; and that you get to choose the pieces (using right-inlet). Whereas [adc~], for example, is also a piecewise-constant interpolator (in hardware or emulated), but all the pieces are identical in "width" (duration), that is, 1 sample.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
Mathieu, Jonathan and Joao,
Thanks for your replies.
I think your words sum up to give me an idea of the current situation with the mentioned docs.
If by any chance someone can explain the historical background, I think I and others as well would be interested in learning..
-- David Shimamoto
On Sat, 26 Sep 2009, PSPunch wrote:
Perhaps I should have made clear that I was not sure how [print~] and [samphold~] would even be considered to go under filters.
No, no, you were clear enough, I just meant that those aren't the only things that don't fit where they've been put, and that the problem is much bigger than that. (e.g. afaik, in Signal Theory, [rpole~] is not a "filter", though it still is peripherally related to filters; otoh there might be other Signal Theorists using different definitions or namings).
Is [samphold~] also often used in building filters?
I don't know... but it isn't "filtering" because what you can get out of it can have a richer spectrum than the original (left-inlet input), and it isn't "linear" either, or "quad", or whatever... it doesn't fit the filter theory much... and I don't see how using it anywhere inside an abstraction can not prevent the abstraction to be a linear filter or quad filter...!
According to my num.analysis book, [samphold~] would be called a "piecewise-constant interpolator", with the warning that "constant interpolator" is somewhat a contradiction of terms; and that you get to choose the pieces (using right-inlet). Whereas [adc~], for example, is also a piecewise-constant interpolator (in hardware or emulated), but all the pieces are identical in "width" (duration), that is, 1 sample.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
On Sun, 27 Sep 2009, PSPunch wrote:
If by any chance someone can explain the historical background, I think I and others as well would be interested in learning..
The historical background of... what? The classification used in Pd itself?
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
No, I know exactly what the relevancy is, I just don't enjoy it. First, a person tells himself/herself «it would be better if there were categories». Then the person looks for characteristic features of the elements to be categorised, so that categories can be made. Those
features have to be easy to think about. Turns out that one of the easiest
features to think about in this case, are things like: where you first learned the basic concept of each object class. It's a kind of microcosm of the whole job-title social structure. Let me give an example.
yes, that's a very good example of category-building. if it's simple,
everyone is going to remember it.
[lop~] is not an operation you learn in elementary-school or high-school math, therefore it doesn't fit in MATH. It doubly doesn't fit in math, because it isn't taught in a Math Department. A Math Department is a social structure that concentrates on any math concept that doesn't
belong to any other discipline already, because if Electrical Engineers already occupy the [lop~] land, it's not only redundant for Math Departments to claim it, it also would make Mathematicians look like Electrical Engineers. So not only [lop~] is not part of Math Depts, but a bunch of related topics are just on the border, so they get lumped into a course called Applied Math, which is all made of pure theory, it's just a form
of discrimination against kinds of Math that are too much in use by other departments. Meanwhile, Electrical Engineers would say that [lop~] is math, except when they get distracted by a category system. But most of all, for music students, [+~] is true math, whereas [lop~] is something magical and not math, because [lop~] is not part of what they learnt in courses labelled as «math» before, so it looks a lot more «audiosome»
than +~ does. This is a summary. The actual situation is more complicated.
ahhh, aren't you going too far? for what is [lop~] mostly used, for math
or as a filter? if as a filter, then why categorize it as a math external?
following your radical assumption, there would be only 3 categories, math
(because 50% of the externals involve some math), dataflow (because the
other 50% of externals route/handle messages), and data structures
(because 0,0001% of objects deal with them). do you think it would be
easier for the "general user" (whatever that is) to find an object
following those guidelines, than looking for "filters", "math" etc?
look at max/msp. did those categories (which are the same as here, but
more detailed) helped or prevented people from using it?
taking your example, if we put a quizz on the list asking if [lop~] is a
math of filter object, what do you think most people will choose?
I'm not completely against categories... I'm trying very hard to make
good categorisations, because it's hard for me to find a categorisation that I can take seriously, and I'm trying to find one.
what you mean with "good" or "take seriously"? you won't want to use
[lop~] if someone thinks that it's more used as a filter as a math object?
At this point, I don't expect Pd's category list to change at all, so, depending on what it is that you're doing, it may be better to just go with Pd's categories, if you have any advantage in following Pd's categories.
I think a major overhaul of Pd's categories list (pd-van + pd-ext) is
necessary. but that will only happen if people really want to discuss it
and get organized. otherwise it will be only more characters traded around
on e-mails.
On Sat, 26 Sep 2009, João Pais wrote:
Those features have to be easy to think about. Turns out that one of the easiest features to think about in this case, are things like: where you first learned the basic concept of each object class.
yes, that's a very good example of category-building. if it's simple, everyone is going to remember it.
What I mean is that using those categories does not much except reinforcing stereotypes that are just artefacts of how things were learned by certain groups of people, at the expense of not just everybody who didn't learn it like that, but also everybody who doesn't want it to be grouped like that.
for what is [lop~] mostly used, for math or as a filter?
both... well, if I understand the question.
I definitely can't pick one OR the other.
but then I cannot tell the difference between something "used as math" and something "used as a filter".
following your radical assumption, there would be only 3 categories, math (because 50% of the externals involve some math), dataflow (because the other 50% of externals route/handle messages),
I thought we were only talking about internal classes, not external (but a good categorisation would expand naturally to externals)
no, if there were a dataflow category, it'd cover about 100% of the classes, and then we'd need subcategories, and math could be a subcategory, and it'd cover perhaps 90% of the classes, and then we'd need a lot of subcategories because it's meaningless to have categories that contain nearly 100% of the items; so instead we'd start from scratch with other ideas.
i exposed some of those other ideas in the mail you replied to, but you skipped over it then told me I'd want only a 3-category system that I don't want either.
do you think it would be easier for the "general user" (whatever that is)
You're not a general user and neither I am. In a community as disparate as pd, there may not be any such thing. But no matter what, there are always people willing to claim themselves as more normal than others.
look at max/msp.
how do I do that?
did those categories (which are the same as here, but more detailed) helped or prevented people from using it?
is that the only thing that matters?
if we put a quizz on the list asking if [lop~] is a math of filter object, what do you think most people will choose?
most people will choose to skip the question.
you won't want to use [lop~] if someone thinks that it's more used as a filter as a math object?
What's this kind of strawman...?
Stalin wore clothes, so, of course, decent people shouldn't. ;)
Really, I meant something more like: I could be removing category names from my copy of 00.INTRO.txt... but I don't think I'd even bother with it. In any case, category systems don't tend to affect my opinion of whatever is put in the category.
I think a major overhaul of Pd's categories list (pd-van + pd-ext) is necessary.
right.
but that will only happen if people really want to discuss it and get organized. otherwise it will be only more characters traded around on e-mails.
right.
...
BTW: if you look up the definition of "Filter" in Signal Theory, you find out that "Filter" is a word used to refer to certain kind of Math functions, that may remove or leave alone any given frequency, but will never boost any frequency. This definition rules out [rpole~] from that category, because the output of [rpole~] can be a bunch of dB above the strength of the input signal.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
On Sat, Sep 26, 2009 at 5:48 PM, Mathieu Bouchard matju@artengine.cawrote:
On Sat, 26 Sep 2009, João Pais wrote:
Those features have to be easy to think about. Turns out that one of the
easiest features to think about in this case, are things like: where you first learned the basic concept of each object class.
yes, that's a very good example of category-building. if it's simple, everyone is going to remember it.
What I mean is that using those categories does not much except reinforcing stereotypes that are just artefacts of how things were learned by certain groups of people, at the expense of not just everybody who didn't learn it like that, but also everybody who doesn't want it to be grouped like that.
There is still some good in this world as categories 1.) are an optional startup plugin (and believe me there will be more startup plugins to piss you off, that's why they are optional) 2.) they are made up of nested lists which even your dog/cat is able to edit in case you don't feel to ;o) This is how they look: *imaging {{gem particles} {gem manipulators} {gem pixes} wrapper particles automata processing}*
On Sat, 26 Sep 2009, András Murányi wrote:
There is still some good in this world as categories 1.) are an optional startup plugin (and believe me there will be more startup plugins to piss you off, that's why they are optional)
yeah, exactly :)
it would definitely solve some social problems if category systems are not considered part of the object-classes themselves. it will save a lot of energy and anger, for better causes.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801
-
András Murányi -
João Pais -
Mathieu Bouchard -
PSPunch