In article .com,
"La Puce" writes:
| On 4 Mar, 17:19, K wrote:
|
| What really frightens me is the whole DNA stuff. The few results I have
| seen reported of that seem to overturn everything I have learnt. Since I
| am older and grumpier than you, I am strongly tempted just to ignore the
| whole thing!
|
| I had jumped when I heard that the London Plane's DNA were closer to
| the Lotus flowers than any British trees.

Well, maybe. I have chased up a few such claims, and I have generally
been disgusted with the academic standard of the papers. Almost all
have based their categorisations on a small subset of characteristics,
and have not justified their choice. A great many of them have happily
quoted analyses that demonstrated two incompatible classifications,
each with 90% probability of being right!

There was an interesting paper I saw when I was working on theoretical
taxonomy (40 years ago, so don't ask for a reference!) that showed that
you could get almost arbitrary, 'significant' classifications from a
random, homogeneous collection of data if you had enough dimensions.
I convinced myself that, with more measurements than items, you would
inevitably get some subsets that produced totally bogus, but highly
'significant' classifications.

And, with DNA data, you are talking tens of thousands of genes, and
hundreds of millions of codons, at least. So my advice is NOT to
believe any such claims (nor to disbelieve them), until they have been
established wisdom for at least 3-4 decades.

| We're all made of stars, as well.

Well, that too. But don't get me started on cosmology - it makes
Swift look positively understated.


Regards,
Nick Maclaren.