>> If the two reconstructions are both possible, then neither is
> convincing.
>
> Even PIE suffers from this. The consonants ... the Laryngeal
A very different type of uncertainty. The correspondences between attested
IE languages, which we signal by the symbol *h1 or *dH or whatever, are
largely beyond dispute in PIE. The disagreements you mention are over the
nature of the original phoneme. That is why PIE is convincing.
What I see with nostratic is a lack of agreement about which words, or which
attested phonemes, correspond. That is why it is not convincing.
Peter