It is not too bad if you have a new website built using these methods but if your website is old it could be a major factor which could lose rankings!

The thing I'd be most cautious of is if you have infinitely crawlable variables that never return a 404, or 301 redirects that direct any non-existent URLs to a working page on your site, instead of a 404.

I agree, 301s are better than canonical meta tags, but they also require more work to setup and keep up.

Search engines have long been able to make sense out of pages that are made up of variables in the url, this is not new news, however from an SEO perspective this is less than ideal. The canonical tag is a band aid at best, but possibly the best solution to pages that are hit by feeds, marketing campaigns or a poor CMS.

Search engines aren't saying bring back variables but rather that it's no longer as detrimental as it has been in the past.

I would say that a 301 redirect is the only way to ensure this, and is in fact far more reliable than rel=canonical. I'm not saying don't use the canonical tag, but also ensure you're using permanent redirects wherever possible to avoid duplicate content issues and preserve link equity. This is true, too, because different search engines do not handle rel=canonical in the same manner, or equally as well.

I know, I know – campaign, affiliate and other systems tend to rely upon parameters to track traffic, conversions and revenue/revenue sharing, and removing the parameters from the string means these data are lost. But there are almost always fairly elegant technical solutions by which information in parameters can be recorded before the URL is redirected to the canonical form. The proviso here is Google Analytics, but even then you can allow ?utm parameters for campaign tracking (as Google “gets” utm).