MathML is a failed web standard (or not?)

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

MathML is a failed web standard (or not?)

Paul Topping
Hi,

Peter Krautzberger of MathJax fame, recently posted this on his own blog:

MathML is a failed web standard
https://www.peterkrautzberger.org/0186/

Obviously, he presents some challenges to the MathML standard and its community. I felt that I had to respond:

Response to Peter Krautzberger's "MathML is a failed web standard"
http://bit.ly/1ZLfCF8

I hope this exchange prompts some serious dialog.

Paul Topping

Design Science, Inc.
"How Science Communicates"
Makers of MathType, MathFlow, MathPlayer, Equation Editor
http://www.dessci.com




Reply | Threaded
Open this post in threaded view
|

Re: MathML is a failed web standard (or not?)

peter murray-rust
I write as a chemist who has tried to do the same thing with Chemistry (CML, Chemical Markup Language). I have been inspired by what I see as the success of MathML and do not regard it as a failure. I am particularly interested in Content MathML as computable maths.

The reality seems to be that it takes a generation for many of these ideas to be implemented. in 1998 SVG seemed to be the obvious way of doing graphics, but after 5 years it looked close to death. After 15 years it's become universal.

CML is used by a small number of enthusiasts. The chemical software manufacturers don't care because they only care about the pharma industry and instruments. So we strugle on with a number of ad hoc broken representations of chemistry, which are still primarily graphical. There is almost no chemistry for blind people.

The real problem is semantics. At the moment the world doesn't care. They will have to in the future. IoT demands semantics. You cannot compute pictures. Binding semantics to maths and chemistry is hard but it will have to come. I'd guess that people will need semantic math in 5 years and chemistry in 15.

If you let the world be driven by browser manufacturers and publishers you will get a sighted-human vision of maths and science. The IoT won't need browsers.

It WILL need semantic maths.


On Fri, Apr 1, 2016 at 11:10 PM, Paul Topping <[hidden email]> wrote:
Hi,

Peter Krautzberger of MathJax fame, recently posted this on his own blog:

MathML is a failed web standard
https://www.peterkrautzberger.org/0186/

Obviously, he presents some challenges to the MathML standard and its community. I felt that I had to respond:

Response to Peter Krautzberger's "MathML is a failed web standard"
http://bit.ly/1ZLfCF8

I hope this exchange prompts some serious dialog.

Paul Topping

Design Science, Inc.
"How Science Communicates"
Makers of MathType, MathFlow, MathPlayer, Equation Editor
http://www.dessci.com







--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
+44-1223-763069
Reply | Threaded
Open this post in threaded view
|

Re: MathML is a failed web standard (or not?)

Andrew Robbins-2
Dear MathML Subscribers,

I must admit that I have not read every word of both posts, but I already know what this is about, because I have already encountered similar issues, with both Presentation and Content. I'm not too concerned with Presentation, because MathJax does an excellent job at that. What I am concerned with is Content (i.e. Semantics), and to quote the original article:

    "Content MathML is just not relevant." -- Peter Krautzberger

I have been writing a set of tools for trans-language compilation for about 5 years now, (freely available at https://github.com/andydude/droxtools), and the only system I've found that is is open, non-commercial, and easily extensible for representing arbitrary concepts from every programming language ever invented, is Content MathML. This is the opposite of "not relevant", and Paul Topping failed to address this.

In my humble opinion, the reason why MathML has failed isn't because of Content MathML, it's because of Presentation MathML, and it's not because it isn't accurate, or because it doesn't look good, it's because people prefer TeX over angle brackets. MathJax provides people the ability to show the same beautiful math expressions on web pages, that Presentation MathML promised but with many fewer keystrokes.

I don't care about angle brackets. I don't care about superscripts. The only thing that interests me with regards to Content MathML, is the fact that it is, at a fundamental level, a LISP where symbols are selected from URI/RDF/XML/MathML namespaces. Granted, OpenMath/MathML namespaces are naturally defined to be equivalent, but to apply to XML/QNames as well, then you need a QName to URI mapping. I've seen two of these, the "{NS}NAME" method (think Java/Ruby/Python) and the "NS::NAME" method (think JavaScript/E4X), the first one fails to produce a valid URI, but the second method does produce a valid URI, so that's what I've been using in my tools. The point is that URIs are already a carefully controlled resource, and so they are much more open than LISP's traditional filesystem based package system, or any other system I've seen.

Just in the interest of full disclosure, there are closed, commercial systems out there that do trans-language compilation, like the kind I'm currently developing. https://www.semanticdesigns.com/ is an example of a corporation involved in such a business. But I whole-heartedly believe that the future of open-source software depends on having such tools available as open source tools. This is starting to sound like a rant, so I will stop it here.

Actually, I changed my mind. I still have to have a Content vs. Presentation debate. As I said earlier, I agree that Presentation MathML has failed, but that's because it's a failed viewpoint. Math isn't symbols, it's semantics. From the beginning, MathML should have been about Content, not Presentation. I think if we had focused on Content all along, then we probably wouldn't be having this conversation now.

Regards,
Andrew Robbins


On Fri, Apr 1, 2016 at 7:21 PM, Peter Murray-Rust <[hidden email]> wrote:
I write as a chemist who has tried to do the same thing with Chemistry (CML, Chemical Markup Language). I have been inspired by what I see as the success of MathML and do not regard it as a failure. I am particularly interested in Content MathML as computable maths.

The reality seems to be that it takes a generation for many of these ideas to be implemented. in 1998 SVG seemed to be the obvious way of doing graphics, but after 5 years it looked close to death. After 15 years it's become universal.

CML is used by a small number of enthusiasts. The chemical software manufacturers don't care because they only care about the pharma industry and instruments. So we strugle on with a number of ad hoc broken representations of chemistry, which are still primarily graphical. There is almost no chemistry for blind people.

The real problem is semantics. At the moment the world doesn't care. They will have to in the future. IoT demands semantics. You cannot compute pictures. Binding semantics to maths and chemistry is hard but it will have to come. I'd guess that people will need semantic math in 5 years and chemistry in 15.

If you let the world be driven by browser manufacturers and publishers you will get a sighted-human vision of maths and science. The IoT won't need browsers.

It WILL need semantic maths.


On Fri, Apr 1, 2016 at 11:10 PM, Paul Topping <[hidden email]> wrote:
Hi,

Peter Krautzberger of MathJax fame, recently posted this on his own blog:

MathML is a failed web standard
https://www.peterkrautzberger.org/0186/

Obviously, he presents some challenges to the MathML standard and its community. I felt that I had to respond:

Response to Peter Krautzberger's "MathML is a failed web standard"
http://bit.ly/1ZLfCF8

I hope this exchange prompts some serious dialog.

Paul Topping

Design Science, Inc.
"How Science Communicates"
Makers of MathType, MathFlow, MathPlayer, Equation Editor
http://www.dessci.com







--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
<a href="tel:%2B44-1223-763069" value="+441223763069" target="_blank">+44-1223-763069

Reply | Threaded
Open this post in threaded view
|

RE: MathML is a failed web standard (or not?)

Paul Topping

Hi Andrew,

 

You mention that you are concerned with the number of keystrokes required to enter MathML. MathML (both Content and Presentation) was never intended to be typed by humans entering math into a computer. MathML is only a computer representation, an internal format to be written and read by software. If you prefer to type your math using TeX, that’s not a strike against MathML.

 

As far as Content vs Presentation is concerned, it sounds like you are correctly choosing Content MathML for your purposes. Nothing wrong with that. However, that is no reason to claim that Presentation MathML has “failed”. When you say “Math isn’t symbols”, you must admit that math is indeed symbols at a certain level – that of presentation.

 

Paul

 

From: Andrew Robbins [mailto:[hidden email]]
Sent: Friday, April 1, 2016 6:42 PM
To: Peter Murray-Rust <[hidden email]>
Cc: Paul Topping <[hidden email]>; [hidden email]
Subject: Re: MathML is a failed web standard (or not?)

 

Dear MathML Subscribers,

 

I must admit that I have not read every word of both posts, but I already know what this is about, because I have already encountered similar issues, with both Presentation and Content. I'm not too concerned with Presentation, because MathJax does an excellent job at that. What I am concerned with is Content (i.e. Semantics), and to quote the original article:

 

    "Content MathML is just not relevant." -- Peter Krautzberger

 

I have been writing a set of tools for trans-language compilation for about 5 years now, (freely available at https://github.com/andydude/droxtools), and the only system I've found that is is open, non-commercial, and easily extensible for representing arbitrary concepts from every programming language ever invented, is Content MathML. This is the opposite of "not relevant", and Paul Topping failed to address this.

 

In my humble opinion, the reason why MathML has failed isn't because of Content MathML, it's because of Presentation MathML, and it's not because it isn't accurate, or because it doesn't look good, it's because people prefer TeX over angle brackets. MathJax provides people the ability to show the same beautiful math expressions on web pages, that Presentation MathML promised but with many fewer keystrokes.

 

I don't care about angle brackets. I don't care about superscripts. The only thing that interests me with regards to Content MathML, is the fact that it is, at a fundamental level, a LISP where symbols are selected from URI/RDF/XML/MathML namespaces. Granted, OpenMath/MathML namespaces are naturally defined to be equivalent, but to apply to XML/QNames as well, then you need a QName to URI mapping. I've seen two of these, the "{NS}NAME" method (think Java/Ruby/Python) and the "NS::NAME" method (think JavaScript/E4X), the first one fails to produce a valid URI, but the second method does produce a valid URI, so that's what I've been using in my tools. The point is that URIs are already a carefully controlled resource, and so they are much more open than LISP's traditional filesystem based package system, or any other system I've seen.

 

Just in the interest of full disclosure, there are closed, commercial systems out there that do trans-language compilation, like the kind I'm currently developing. https://www.semanticdesigns.com/ is an example of a corporation involved in such a business. But I whole-heartedly believe that the future of open-source software depends on having such tools available as open source tools. This is starting to sound like a rant, so I will stop it here.

 

Actually, I changed my mind. I still have to have a Content vs. Presentation debate. As I said earlier, I agree that Presentation MathML has failed, but that's because it's a failed viewpoint. Math isn't symbols, it's semantics. From the beginning, MathML should have been about Content, not Presentation. I think if we had focused on Content all along, then we probably wouldn't be having this conversation now.

 

Regards,

Andrew Robbins

 

 

On Fri, Apr 1, 2016 at 7:21 PM, Peter Murray-Rust <[hidden email]> wrote:

I write as a chemist who has tried to do the same thing with Chemistry (CML, Chemical Markup Language). I have been inspired by what I see as the success of MathML and do not regard it as a failure. I am particularly interested in Content MathML as computable maths.

The reality seems to be that it takes a generation for many of these ideas to be implemented. in 1998 SVG seemed to be the obvious way of doing graphics, but after 5 years it looked close to death. After 15 years it's become universal.

CML is used by a small number of enthusiasts. The chemical software manufacturers don't care because they only care about the pharma industry and instruments. So we strugle on with a number of ad hoc broken representations of chemistry, which are still primarily graphical. There is almost no chemistry for blind people.

The real problem is semantics. At the moment the world doesn't care. They will have to in the future. IoT demands semantics. You cannot compute pictures. Binding semantics to maths and chemistry is hard but it will have to come. I'd guess that people will need semantic math in 5 years and chemistry in 15.

If you let the world be driven by browser manufacturers and publishers you will get a sighted-human vision of maths and science. The IoT won't need browsers.

It WILL need semantic maths.

 

 

On Fri, Apr 1, 2016 at 11:10 PM, Paul Topping <[hidden email]> wrote:

Hi,

Peter Krautzberger of MathJax fame, recently posted this on his own blog:

MathML is a failed web standard
https://www.peterkrautzberger.org/0186/

Obviously, he presents some challenges to the MathML standard and its community. I felt that I had to respond:

Response to Peter Krautzberger's "MathML is a failed web standard"
http://bit.ly/1ZLfCF8

I hope this exchange prompts some serious dialog.

Paul Topping

Design Science, Inc.
"How Science Communicates"
Makers of MathType, MathFlow, MathPlayer, Equation Editor
http://www.dessci.com





--

Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
<a href="tel:%2B44-1223-763069" target="_blank">+44-1223-763069

 

Reply | Threaded
Open this post in threaded view
|

Re: MathML is a failed web standard (or not?)

Peter Krautzberger
In reply to this post by Andrew Robbins-2
Hi Peter and Andrew,

Thanks for those interesting statements. I'm not sure how they relate to what I wrote (please let me know if I missed something) but I appreciate having the opportunity to read them.

Regards,
Peter.

On Sat, Apr 2, 2016 at 3:42 AM, Andrew Robbins <[hidden email]> wrote:
Dear MathML Subscribers,

I must admit that I have not read every word of both posts, but I already know what this is about, because I have already encountered similar issues, with both Presentation and Content. I'm not too concerned with Presentation, because MathJax does an excellent job at that. What I am concerned with is Content (i.e. Semantics), and to quote the original article:

    "Content MathML is just not relevant." -- Peter Krautzberger

I have been writing a set of tools for trans-language compilation for about 5 years now, (freely available at https://github.com/andydude/droxtools), and the only system I've found that is is open, non-commercial, and easily extensible for representing arbitrary concepts from every programming language ever invented, is Content MathML. This is the opposite of "not relevant", and Paul Topping failed to address this.

In my humble opinion, the reason why MathML has failed isn't because of Content MathML, it's because of Presentation MathML, and it's not because it isn't accurate, or because it doesn't look good, it's because people prefer TeX over angle brackets. MathJax provides people the ability to show the same beautiful math expressions on web pages, that Presentation MathML promised but with many fewer keystrokes.

I don't care about angle brackets. I don't care about superscripts. The only thing that interests me with regards to Content MathML, is the fact that it is, at a fundamental level, a LISP where symbols are selected from URI/RDF/XML/MathML namespaces. Granted, OpenMath/MathML namespaces are naturally defined to be equivalent, but to apply to XML/QNames as well, then you need a QName to URI mapping. I've seen two of these, the "{NS}NAME" method (think Java/Ruby/Python) and the "NS::NAME" method (think JavaScript/E4X), the first one fails to produce a valid URI, but the second method does produce a valid URI, so that's what I've been using in my tools. The point is that URIs are already a carefully controlled resource, and so they are much more open than LISP's traditional filesystem based package system, or any other system I've seen.

Just in the interest of full disclosure, there are closed, commercial systems out there that do trans-language compilation, like the kind I'm currently developing. https://www.semanticdesigns.com/ is an example of a corporation involved in such a business. But I whole-heartedly believe that the future of open-source software depends on having such tools available as open source tools. This is starting to sound like a rant, so I will stop it here.

Actually, I changed my mind. I still have to have a Content vs. Presentation debate. As I said earlier, I agree that Presentation MathML has failed, but that's because it's a failed viewpoint. Math isn't symbols, it's semantics. From the beginning, MathML should have been about Content, not Presentation. I think if we had focused on Content all along, then we probably wouldn't be having this conversation now.

Regards,
Andrew Robbins


On Fri, Apr 1, 2016 at 7:21 PM, Peter Murray-Rust <[hidden email]> wrote:
I write as a chemist who has tried to do the same thing with Chemistry (CML, Chemical Markup Language). I have been inspired by what I see as the success of MathML and do not regard it as a failure. I am particularly interested in Content MathML as computable maths.

The reality seems to be that it takes a generation for many of these ideas to be implemented. in 1998 SVG seemed to be the obvious way of doing graphics, but after 5 years it looked close to death. After 15 years it's become universal.

CML is used by a small number of enthusiasts. The chemical software manufacturers don't care because they only care about the pharma industry and instruments. So we strugle on with a number of ad hoc broken representations of chemistry, which are still primarily graphical. There is almost no chemistry for blind people.

The real problem is semantics. At the moment the world doesn't care. They will have to in the future. IoT demands semantics. You cannot compute pictures. Binding semantics to maths and chemistry is hard but it will have to come. I'd guess that people will need semantic math in 5 years and chemistry in 15.

If you let the world be driven by browser manufacturers and publishers you will get a sighted-human vision of maths and science. The IoT won't need browsers.

It WILL need semantic maths.


On Fri, Apr 1, 2016 at 11:10 PM, Paul Topping <[hidden email]> wrote:
Hi,

Peter Krautzberger of MathJax fame, recently posted this on his own blog:

MathML is a failed web standard
https://www.peterkrautzberger.org/0186/

Obviously, he presents some challenges to the MathML standard and its community. I felt that I had to respond:

Response to Peter Krautzberger's "MathML is a failed web standard"
http://bit.ly/1ZLfCF8

I hope this exchange prompts some serious dialog.

Paul Topping

Design Science, Inc.
"How Science Communicates"
Makers of MathType, MathFlow, MathPlayer, Equation Editor
http://www.dessci.com







--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
<a href="tel:%2B44-1223-763069" value="+441223763069" target="_blank">+44-1223-763069


Reply | Threaded
Open this post in threaded view
|

Re: MathML is a failed web standard (or not?)

William F Hammond
In reply to this post by Andrew Robbins-2
Going back to April 1, Andrew Robbins writes in part:

> In my humble opinion, the reason why MathML has failed
> isn't because of Content MathML, it's because of
> Presentation MathML, and it's not because it isn't
> accurate, or because it doesn't look good, it's because
> people prefer TeX over angle brackets. MathJax provides
> people the ability to show the same beautiful math
> expressions on web pages, that Presentation MathML
> promised but with many fewer keystrokes.

I believe that almost all extant MathML content that I've
seen originates, one way or another, with LaTeX or
LaTeX-like markup.

> I don't care about angle brackets. I don't care about
> superscripts. The only thing that interests me with
> regards to Content MathML, is the fact that it is, at a
> fundamental level, a LISP where symbols are selected from
> URI/RDF/XML/MathML namespaces.  . . .

I think it's rather difficult to generate reliably useful
content MathML from LaTeX markup as commonly seen, for
example, at arXiv.  On the other hand, I believe that adding
a LaTeX package for type declaration of mathematical symbols
would go a long way toward improving this.  Even better
would be the use of a suitable LaTeX profile (such as I
spoke about at TUG 2010 and TUG 2014) with provision for
symbol type declarations.

> . . .  As I said earlier, I agree that Presentation MathML
> has failed, but that's because it's a failed viewpoint.
> Math isn't symbols, it's semantics. From the beginning,
> MathML should have been about Content, not Presentation. I
> think if we had focused on Content all along, then we
> probably wouldn't be having this conversation now.

>From the beginning the development of MathML has had two
tracks with content MathML focused on content interchange
and presentation MathML designed for minimizing the amount
of work required for a web browser to provide a TeX-quality
rendering of mathematical content from an extension of HTML
markup.

I disagree with the assertion that presentation MathML has
failed as a web standard.  It works quite well in Firefox
and other Gecko browsers, and one should not forget W3C's
Amaya.  It is, of course, disappointing that, for the moment
and for most of the time since the beginning of MathML in
the late 1990s, three of the "big four" browsers have not
had native support for MathML.

It's a failing in the market based on crass market thinking.

It was also disappointing that for the period from 2001 (if
not 1995 with the dropping of HTML v 3.0) to 2011 the W3C
banned any form of math content from the media type
"text/html".

Of course, there would be breakage of existing content if
web browsers that now render MathML ceased to do so.

It continues to be disappointing that search engines do not
cover mathematical content well.

The disappointments are not failures, but rather the
result of a world where a relatively small number of
individuals have any interest in mathematics.

Still, native browser rendering of MathML could happen.
Didn't Murray Sargent just say so?  There is no reason
to stop wishing for it.

                              -- Bill