The unambigious definition would be a contiguous landmass surrounded by water at sealevel, which would be in descending order of area

Afro-Eurasia *

Americas *

Antarctica

Australia landmass

Greenland

* This of course depending on how rivers and man-made canals of Suez, Panama, and other canals like Kiel, or ones linking rivers, are treated, there could be arguments that Suez should be included (no locks), but Panama shouldn't be, or all should be, or none should be, or Suez and Panama should be, etc

"So the reason greenland being an island is due to definition then?"

Isn't everything, including criteria?

No, it's the difference between normative and descriptive categorizations.

But deciding that some concept A is defined by criterion / set-of-criteria B is itself normative, a definition by fiat.

> Which makes the concept of an island moot, as it's driven via definition, rather than criteria.

So all concepts that are "driven via definition, rather than criteria" are moot?

Isn't that... All of them?

Countries are defined by convention as well; that doesn’t make them a moot concept.

That aside, there is actually a rather good criteria-based definition for continents vs. islands: https://news.ycombinator.com/item?id=35568455

What the difference between definition and criteria?

a criteria can be applied to something to find out if it fits - a definition is by fiat (without the use of any criteria).