A couple of (further) points:
- There is no further definition of rock in UNCLOS.
- There isn't much precedent before the Philippines-China case on the interpretation of those (article 121(3)) provisions.
I say this based on a (South African) PhD thesis on the topic. And also based on that:
There was a prior case (Romania v Ukraine on the Serpent Island in the Black Sea) in which the parties argued extensively around the article 121 definition, but the ICJ ignored those arguments entirely, and ruled instead that the island/rock in question could not be used to alter the provisional equidistance line between the two countries.
And for some details on the Philippines-China case, the thesis has a pretty extensive analysis. The most relevant bits seem to be:
The Tribunal was of the view that “habitation” also implies that the insular feature must
be inhabited by a group or community. Although Article 121(3) does not specify any specific
number of people, the habitation of an insular feature by a sole person would not typically be
considered human habitation in the ordinary sense.
[...]
The Tribunal was of the view
that in order for an economic activity to be classified as the economic life of an insular feature,
the resources on which the economic activity centres must be local as opposed to imported and
must also be to the benefit of the economic activity in question. Thus, any form of economic
activity that relies on the injection of external resources for its continuation does not fall within the
scope of “an economic life their own”. Economic activity of this kind would not constitute the insular
feature’s own economic life, but rather an economic life which is reliant on outside support. [...] economic activity derived
from an insular feature’s territorial sea may constitute the economic life of the feature only if it is
connected to the feature itself, be it via the local population or via other mediums. This means
that the exploitation of the territorial sea by distant fishermen or extractive enterprises which
operate in the territorial sea of an insular feature, but make no use of the actual feature, will not
be sufficient in bestowing the insular feature with an economic life of its own.
So I think the decision (on this issue) rested on the lack of connection between the group living on the island and the economic activity (around the island).
Of course, Taiwan (and China) disagreed with the court's conclusion.
As an aside, according to a Washington Post article, the fact that UNCLOS has no explicit definition of "rock" is the basis for Japan's claim to Okinotori:
Relevant to the issue is the definition of “island” and “rock.” UNCLOS stipulates that “an island is a naturally formed area of land, surrounded by water, which is above water at high tide” [Article 121 (1)]. The law does not clearly define what a rock is, specifying only that “[r]ocks which cannot sustain human habitation or economic life of their own shall have no exclusive economic zone or continental shelf” [Article 121 (3)].
So UNCLOS has a big loophole. Japan claims Okinotori is an island, based on Article 121 (1), but also maintains that Article 121 (3) is about rocks and thus is irrelevant because there is no clear definition of “rock.” This created a useful excuse for Japan and other countries (including China, in the South China Sea) for their island-building projects.
One point is clear, though: An island can be the base point of establishing an EEZ, but a rock cannot. Despite the fact that scholars of international law, citing Article 60 (8), have long argued that artificially built islands do not generate an EEZ, the Japanese government has been engaged in island-building on Okinotori for decades.