Google has provided this for years with the "site:" feature (e.g. search for "lists site:python.org"). In addition, Google uses its own results, since as the article correctly mentions "most site’s internal search engines suck".

> Google has provided this for years with the "site:" feature

Which is something completely different, and is also something present in DuckDuckGo.

But chrome and firefox provide it for years, without giving up your queries to a 3rd party search engine.

In this context, all search engines are "3rd party".

If you want to search amazon, or the python website, why give your query to ddg? Especially if you don't have to when you use browser keywords.

In a word: curation.

Amazon is probably always going to be the most relevant search engine for their site, but the current choice for python searches may not be.

An example similar to python but that I have first hand experience with is Perl's CPAN searches. There are two search engines for CPAN, http://search.cpan.org and http://metacpan.org. DDG's !cpan originally defaulted to the first one, but at the time it's results were not as accurate or informative as the second one. Upon appeal the DDG people switched the default to http://metacpan.org providing everyone who used the !cpan results with what they felt was a better result.

"site:" query in Google searches Google index and filters results by domain name, while DDG actually delegates to the actual site and uses the site's search engine, which very rarely is of an acceptable quality.