For centuries careful researchers and scholars have trusted in the stability of citations. Once a work was cited it was always, in principle, available to be checked and verified. Those centuries of reliance on the stability of citations is at an end. Two decades of academic literature now relies on dead links and vanished sources as writers mistakenly believed URLs were as permanent as cited books.
The essential method of linking to sources, the citation, is imperiled in the age of the Internet by the constant flux of the Internet. The growing use of citations to URLs is a perilous practice, since the not only do individual URLs change, but documents and even entire websites vanish all the time. (One little used quick fix is archiving a webpage and then providing the URL of the archive page.) Now even searching for a specific URL is no guarantee that you will not be re-routed elsewhere.
Moreover, outside curated databases, there is little assurance of reproducible searches. The same problems that make citations unstable, make searches not reproducible: Useful search results produced on Monday morning might be gone without a trace in the afternoon.
Data curation in databases means inclusion of useful information, while data curation by search engines means exclusion of information by unknowable and evolving “algorithms”. By definition, curated databases have stability of citation and reproducible searches, while online search engines do not.