RE: agenda+ SVG 2 Features and Approach

On  Sunday, May 08, 2011 1:09 PM

Doug Schepers writes: 

>Yes, this is a good idea; the SVG WG agreed with the idea of a poll.  We 
>definitely want to get the opinions of the larger developer community, 
>so we want a wider reach than just this list.

>W3C is setting up some improved forum-based polling (where people can 
>post ideas and others can mod them up or down), but I'm not sure when 
>that will be put into effect, nor whether it will be suitable for this 
>kind of idea collection.  (It might work well, I'm just not sure yet.)

Hi Doug and others, 

The word "poll" comes with a lot of connotations for me, many of which raise methodological flags. Polls convey this sense that outcomes are countably many (in the sense of Cantor) and that each outcome is categorizable, on at least a nominal scale. The term "discussion" is not so focused as "poll," but perhaps less actionable. Maybe something in between is best, and maybe that's what Doug is saying.

I think that a proper methodology for making any information from such polls "usable" might include more than one phase. 

For example, in phase one (which is sort of ongoing) several people (rather more expert in nature, we may presume) will respond with lists of various sizes. We might call this a "brainstorming phase."

During this phase, some folks (like me) might not understand quite what others are actually saying (I use as examples [1] by Olaf Hoffmann and [2] by Charles Pritchard, where I get the sense that with some effort and thought on my part I probably could understand, and that once that happened, I would be likely to agree, but in the meantime, I think a few back and forths may be needed for each). Perhaps, Doug, that is what you mean by " forum-based polling (where people can 
>post ideas and others can mod them up or down)" -- Modding is not a verb I know, nor is "forum based polling" something I'm familiar with. 

Rather than just accumulating a list of m features (k of which would form a spanning or generating set for the other m minus k) from N respondents (where N is hopefully very large) and then having the WG review and decide which ones the implementers wish to implement, it might be nice to slow things down just a bit:

1. Cast a fairly broad net for recommended features and functionalities
2. Discuss them so that a list of mutually exclusive features could actually be constructed using examples and language that would be broadly comprehensible. Then we might all at least come to know what each other is saying. Maybe the SVG-IG would be a good place to have those discussions so the SVG WG archives don't become so clogged as have the HTML WG archives?
3. Determine subsets of the m features that span equivalent functionality sets.  For example, implementing functionalities (A,B, C, and D) might offer something equivalent to (B, E, F, G, H and I) and with less cost to authors but more to implementers; or (X,Y, Z) may carve out low lying fruit in a way that satisfies implementers but forces authors to jump ship and go port Second Life to the web and thence abandon SVG. (I'm just pretending here --the point being that different solutions may appeal to different communities, and some with different costs accordingly.)
4. Following some discussion of use cases relating to a rather broad list of wishes, then one might construct a short answer "poll" in the classical sense to get some sense of what features people actually think they need. Doing so beforehand is likely to under-represent the importance of certain features, since respondents might not know that other alternatives exist, nor of what the benefits of certain approaches might be.

The reasons for methodological considerations being raised are that a) I have this unshakable sense that in the early days of HTML5, the conversations were pre-empted by certain irrepressible momenta, not all of which were interested in democratic input and not all of which were beneficial to the development of the spec, b) good ideas are sometimes lost, forgotten, or squelched without due process c) as someone who has taught and researched creative methodologies, my tolerance for chaos and balkanization in public decision-making is not as developed as it could be -- I'm trying to work on developing more appreciation for chaos, but it seems to require lifetimes.

5. Ultimately the results of such a poll might merely reflect one measure of "authorial interest", and little more. Ideally, a broad discussion (steps 1 through 4 above) within the community about how to do the various things that people want to do would be a far more useful outcome than the results of a poll itself, the methodology of which would inevitably be questionable. (68% of all people agree that poll taking methodology is suspect.) But as the process unfolds (discussions, use cases and opinion surveys), then the Working Group (who ultimately have responsibility for developing a spec that implementers will be willing to implement) would find itself informed by the entire process.

As you know I think SVG has the opportunity to become really big [3],[4] -- bigger than HTML, since its underlying metaphor is closer to the human thought process. As such, I think the process of building it should be a bit both careful and open.

Cheers
David

[1] http://lists.w3.org/Archives/Public/www-svg/2011May/0051.html 
[2] http://lists.w3.org/Archives/Public/www-svg/2011May/0055.html 
[3] http://srufaculty.sru.edu/david.dailey/svgOpen2009/SVgBigTalk.html 
[4] https://www.svgopen.org/2009/papers/26-Why_is_SVG_going_to_be_REALLY_BIG_/ 

Received on Monday, 9 May 2011 00:49:37 UTC