Using speed of polymerase to sequence DNA?

[Jarek Duda]

There are considered some approaches to DNA sequencing while it goes
through some nanoscale hole (with some nanoelectrodes) and so many
orders of magnitude faster than e.g. pyrosequencing.
Unfortunately even theoretical simulations says that identifying bases
this way is already extremely difficult ...
http://pubs.acs.org/doi/abs/10.1021/nl0601076

Maybe we could use nature's ways to read/work with DNA for this
purpose?
For example somehow mount polymerase or ribosome and somehow monitor
its state...

I thought about using speed of process to get information about
currently processed base.
For example DNA polymerase to process succeeding base has to get from
environment corresponding 'nucleotide carrier' (nucleoside
triphosphat) - there are only four of them - we can manipulate their
concentrations.
If we would choose different concentrations for them, there would be
correlations between type of the base and time of its processing - by
watching such many processes we could determine the sequence.

For example - somehow mount polymerase on the cantilever of atomic
force microscope, so that it can 'watch' its speed of DNA processing.
Now use different concentrations of the 'carriers' of nucleotides - so
that the speed of the process depends on the current base.
So there should be correlations between base sequence and forces
observed by the microscope - processing given sequence a few times
this way, we should be able to fully determine base sequence ... many
orders of magnitude faster than using pyrosequencing.
Eventually the movement of polymerase could be watched optically (for
example by attaching to it something producing light, like luciferase
and surround the space with CCD).

There are a few random factors which would have some small influence
on this speed and still it's stochastic process - sometimes it will
quicker catch 'nucleotide carrier' of smaller concentration ... its
why it would be required to process given ssDNA a few times and use
some Bayesian analysis to determine the sequence.
To reduce such effect we could somehow mount straightened DNA and
choose really large differences in concentrations (even like
1:10:100:1000)

Is it doable?
What do you think about such methods of sequencing while going through
a hole?
How to use proteins developed by nature for this process?