Hi! I was wondering if it would be possible to develop a tool for Wikipedia editing which allows (1) numbered lists whose numbering continues throughout the article and (2) an automated way of referring back to earlier examples.

This would be enormously useful for linguistics articles, and an editor at the help desk suggested that it could be useful for some math and science articles as well. Currently, linguistics articles format their examples in ugly and confusing ways, and many rely on text that breaks the fourth wall in order to achieve a minimum standard of clarity ("as we saw in the second sentence example under the third paragraph four subsections ago…") . My suggestion is that editors could write markup like this (or perhaps less cumbersome!):

Example <exref = “donkey” /> shows that donkey anaphora exists. Example <ex ref = “donkeyRC”> uses a relative clause, while Example <ex ref = “donkeyif /> uses a conditional.

  <ex name=“donkey” /> Donkey Anaphora:
     <subex name = “donkeyRC” />Every farmer who owns a donkey beats it.</subex>
     <subex name = “donkeyif” />If a farmer owns a donkey, he beats it.</subex>
  </ex>

Blah blah blah imagine this text runs for several paragraphs, perhaps with a section break in the middle. Now compare the pair shown in <ex ref = “donkey” /> to the completely unrelated ones in <exref = “kangaroo” />.

  <ex name=“kangaroo” /> Counterexample to Antecedent Strengthening:
     <subex name = “kangarootail” />If Kangaroos didn't have tails, they would topple over.</subex>
     <subex name = “kangaroojetpack” /> If Kangaroos didn't have tails, but did have jetpacks, they wouldn't topple over.</subex>
  </ex>

And this would show up as:

Example (1) shows that donkey anaphora exists. Example (1a) uses a relative clause, while Example (1b) uses a conditional.

(1) Donkey Anaphora:

a. Every farmer who owns a donkey beats it.
b. If a farmer owns a donkey, he beats it.

Blah blah blah imagine this text runs for several paragraphs, perhaps with a section break in the middle. Now compare the pair shown in (1) to the completely unrelated ones in (2).

(2) Counterexample to Antecedent Strengthening:

a. If Kangaroos didn't have tails, they would topple over.
b. If Kangaroos didn't have tails but did have jetpacks, they wouldn't topple over

Crucially, the numbers would all update consistently if the page is edited to change the order of the examples, or to stick in intervening examples. I imagine that something like this could be built based on whatever currently handles reference numbering, and if any further inspiration is needed, various LaTeX packages (e.g. linguex) have this functionality.