[Greasemonkey] Question: what would happen if GreasemonkeysenttheID of every applicable user script with each request?

[Matthew Gertner]

Paul,

> 1.	If you're a web developer who is interested in whether or not GM is
> being used on your site, you'll watch the repository.  You'll want to
> know.  You'll find out.
> 
> 2.	If you're interested in fixing the kinds of issues GM is used in the
> "broken" cases, you're probably already working on them in response to
> prior reader feedback, or because you've watched the repository.  For
> example, I went several rounds over the course of a year trying to get
> someone at Audible to even give an intelligent comment on the usuability
> problems of their site.  After several fruitless efforts (blogged
> elsewhere), I gave up and wrote a GM script to fix it, which I also
> forwared to them.  They know about the problem, the fix, and that a
> number of people are using it.  It's just that at some decision-making
> level, they don't care.  Log entries won't change that.

The kind of approach I am talking about would be targeted exactly at web
developers who *don't* have GM on their radar. Despite the success of GM, I
would suspect that this is still the vast majority. Your Audible example
illustrates why this could be effective. Imagine that instead of just
writing to them, you had canvassed the net and found a bunch of other people
who were equally disturbed by their usability issues. Then you all signed a
petition and sent it to them. Don't you think this might have more impact? 

Well, this would be exactly the effect of having GM script information in
loggable form in the request header. First of all, people who are GM-unaware
will be alerted to its existence. Secondly, the scope of usage will be clear
("X% of visitors are viewing our site with this script!"), which would have
much more weight with big companies that a single person sending them an
email, and would be very interesting even to GM-savvy folks who know that a
script exists but have no idea how widely it is used.

> 3.	"with an opt out for sites that try to abuse this" is where another
> arms race begins.  What defines abuse?  And if we declare a site so
> abusive that we don't want to admit to using GM there, what's to stop
> them sniffing it out via Javascript; and we sniff out the sniffing; and
> they sniff out *that*...  And then come the Forrester reports and
> newsbytes about those oppressive GM guys singling out and defaming those
> poor little websites -- if these seems far-fetched, note the recent
> hand-wringing over the mean, nasty, huge Web Standards Project singling
> out poor little Microsoft when discussing the ACID2 CSS tests.

I'm not suggesting that anyone define abuse. If a script author wants it to
be stealthy, he or she adds the appropriate metadatum. Period. Likewise, I'm
not sure that this would be a PR issue for GM. On the contrary, in the face
of the Forrester report (which I finally read and which is shockingly
negative about GM) and inevitable further negative press, it would be
fantastic to be able to demonstrate that the GM community is listening and
proactively coming up with ways to work with site operators who are
concerned about these issues.

> Like so many things, GM is a great example of the value of
> loosely-connected, decentralized adaptation.  List all the complications
> you can think of ahead of time, if we get further entwined with the
> server side of the equation.  Now triple that number, as a rough estimate.

Can you give an example of this? I'm not sure I understand why giving
feedback to the server would increase these complications at all.

Matt