I’ve done a number of Gemini prompts where I’ve asked it to explain, with sources, various topics. And, not infrequently, after each piece of the response which has a corresponding source behind it, Gemini gives me a linkage symbol. And, when you copy and paste it into a blog post, those linkages become a series of numerical footnotes… without the cross-reference to the source. And some of these are duplicated (I asked it whether or not that was true).

Do I really want to bore a non-expert with all of these? Do you really need to know that this snippet came from that source? Of course not – but I did want to make Gemini show its work.

So, I came up with this:

Instead of inserting citation markers throughout the response , provide a source list, including URLs in parentheses (as appropriate), at the end of the response. If multiple citations refer to the same base domain, refer to the base domain instead of the specific page.

It may not be perfect, but it seems to work – I used it on that “nunation” post.

If someone has better ideas, let me know. I’m sure there are equivalent terms to “base domain” – I certainly didn’t want to use “TLD”, although I’m sure I could have said “DNS address for the most specific common URL” or some such.

Note: Gemini refers to these as source markers or citation markers. What other AI engines call them – or how they’re represented in responses – I can’t say. At some point, might I have a shootout with ChatGPT, CoPilot, Claude, etc? Sure… but not today.