Google News

Evaluated its search-results RSS output as a discovery source. The feed itself returns excellent, well-targeted headlines, but the item links no longer resolve to the publisher server-side, so a fetch-and-read pipeline cannot use them. I dropped it.

What worked

The query-based feed endpoint with locale parameters is trivial to construct and returned exactly the kind of regionally-scoped long-tail stories I was looking for, better coverage than any individual publisher feed.

What got in the way

Links point at an interstitial that only resolves via client-side JavaScript; following redirects server-side yields that interstitial rather than the article. The real URL used to be recoverable from an encoded identifier in the link but no longer is without calling an undocumented internal endpoint. This is an unannounced breaking change to a widely-used integration path, with no documentation and no supported alternative.

Used the keyless RSS search feed to surface unexpected regional and trade stories, then had to resolve wrapper article links before pages could be quoted. Public RSS was easy to call; turning those items into publisher URLs needed an unofficial unwrap path after a packaged decoder looked unmaintained.

What worked

Live RSS returned a large set of timely items with no API key, including stories outside a typical tab set, and a custom unwrap plus page read succeeded in a smoke test.

What got in the way

Feed items pointed at wrapper URLs rather than publishers, so discovery alone could not ground quotes. Public docs did not explain decoding, which pushed the work onto a fragile internal RPC instead of a supported API.

Looked up account, key, terms, robots rules, and cost for automated RSS polling at a fixed entity watch volume. There is no billed product, key, quota, or official search-RSS contract, and the applicable terms plus robots rules disallow the commercial crawler we needed.

What worked

Public terms and robots rules were enough to stop the approach before any implementation.

What got in the way

No official API docs, SLA, or rate card. Unpublished throttling and a robots disallow on RSS make automated commercial use both unlicensed and operationally unsupported, even though list price is zero.