whats CURRENTLY the most EFFECTIVE article marketing strategy?

Correct.

It'll get you the juice if your site is the better place, with better cumulative on-page and off-page SEO. This is a longer-term thing: your site will be the better place, with better cumulative on-page and off-page SEO if you always post all your articles on your own site first and have them indexed there before submitting them anywhere else, for all the reasons explained in detail here by so many professional article marketers.

Correct.

And if the other sites on which you're posting are article directories, that's really, really trivial: your site will need comparatively little content, comparatively little history and comparatively little authority to outrank an article directory for its own keywords.

This was true even long before Google's Panda update, and is all the more true, now.

People who always publish their own work on their own sites first typically have absolutely no difficulty at all outranking article directories.

For me, also. Easy if you know how - and that's putting it mildly: it's actually pretty easy for anyone who always posts all their articles on their own site first and has them indexed there before submitting them anywhere else.

Correct.

From an article directory, with its non-context-relevant, PR-0 backlink, it seems to me that fewer and fewer people, these days, "want a backlink". Targeted traffic, on the other hand, is a different and far more beneficial matter.

For me, the key question has always been: when someone finds one of my articles by putting one of its keywords into a search engine as their search engine, which do I want them to find - the copy on my own site ("job done") or the copy in EzineArticles (like most marketers, I lose most of the traffic that I "send" there instead)? So it's a bit of a no-brainer, really.

Not so much, in my view: that would really be only "article directory marketing", not "article marketing" per se.

It's very apparent that most people who fail at "article marketing" tend to fail in more or less the same way, and it's a way far more accurately known as "article directory marketing" and broadly characterised by ...

(i) depending on article directories for traffic and backlinks;
(ii) using "spinning" and/or "mass-submission" software;
(iii) using a "rinse and repeat" model of "article directory marketing";
(iv) writing/submitting large numbers of shorter articles which are sometimes "salesy" in tone and have a prominent "call to action" in the resource box.

You can be pretty sure, whenever you see one of the ever-increasing plethora of threads we regularly have appearing here with titles like "Article Marketing Doesn't Work Any More" or "I'm Not Sure About Article Marketing Any More" (we've all noticed those, right?), that the person starting it off has been doing at least three of the four things mentioned above.
There are reasons for that.

In contrast, most people who succeed at article marketing tend to succeed in their own slightly different ways, a little harder to describe succinctly, but often (I don't say "always", of course) characterised by ...

(a) avoiding all the above;
(b) involving article syndication in their business model;
(c) building an asset-based business, by developing their own virtual properties rather than anyone else's, which produces gradually increasing residual income from work already done.

As Paul ("myob") wisely comments above, syndication works very quickly for some of us, and you don't need many articles at all for it to be effective.

It's a very good and valid method if you can find someone who has the right skill-set to handle it all profitably and productively for you. But if you find such a person, wouldn't they be better off doing it for themselves, rather than for you, and generating their own long-term, residual income? And if they haven't quite worked that out, how confident would you be that they really do have the right skill-set to handle it all for you?

Is this typical after a Google dance?

It's very interesting that you say this is still recommended, even though I thought that this was the very same thing that was frowned upon by Google