It is notable that the opinions expressed in the following links are not as definite as yours.
Link 1
Link 2
Link 3
Link 4
And I like the comment contained in the following where a member of the OSU team hedges on its count. IOW-- "I strongly suspect our numbers are high"
Link 5
"Some researchers are unsettled by the certainty with which the Human Genome Consortium is presenting its lower gene count," said Fred Wright of Ohio State University. "In my view, the final number of genes - when it is known - will lie somewhere between their high of 40,000 and our value of 70,000."
It isn't much more now. We know that DNA codes for proteins and we know that proteins are involved in the expressions of traits. What we don't know is how many proteins are involved per trait - but it's looking like it varies from trait to trait, which makes a lot of common sense. The assertion that it was a one-to-one ratio was really more one of science popularization than real science - it makes it easy to visualize, but it's a major oversimplification to which science popularization is prone. But I don't think any working molecular biologist has been proceeding under that assumption for decades if ever.
The Human Genome Project's actual product was a listing of DNA sequences, and nothing more. There are a number of crude methods of estimating protein expression from this (simply enumerating stop codons is one) but no precise one, and no method at all of estimating traits, hence genes. Anyone who ever said it could do that was either misunderstanding the process or deliberately misrepresenting it. Without pointing any fingers I will observe that science journalists are prone to the first and scientists hungry for grant money to the second. Of course, the same is true of "debunkers."