Actually, close, but only close.

Basically, your sensor counts photons in each sensor cell by
converting[1] them to electrons.  Each cell, at the end of
exposure, will then pass the collected electrons[2] and each
charge amount will then be multiplied by a factor before run
through the A/D converter (which spits out the numbers for
the RAW format).  If you record in JPEG, further work on the
digital data is done, obviously.

Changing the ISO setting will only change the multiplier
setting (which may well be 0.05 (i.e. "divide by 20") or 30
(i.e. multiply by 30)).

The sensitivity of the sensor _as such_ is unchanged, it'll
still collect 500 photons at ISO 100 as on ISO 128.000, given
the  same brightness, aperture and exposure time.  However,
whether that represents a very dark or a very bright part of
the image is determined by the ISO setting.

The reason for the ISO setting is that the A/D converters
only have a limited number of bits (from 10 for compact
cameras to 14 or even 16 for some expensive backends).
Additionally the electron count == charge is linear (twice
as bright == double count) and absolute.

But our eye is comparative (dark vs light, not absolute values)
and, being able both to see over huge absolute ranges (LV 16+
(white/reflective in full sun) to LV -5 (half moon lit objects s
at night)) and high contrast scenes (like direkt sun and objects
in the shadow at the same time).

What you do then is dampen (or boost) all charges, so that
(hopefully) th converter will get near, but *never* over 100%
at the largest charges and (hopefully) still enough bits engaged
for nice gradual changes for the darkest parts.  As you cannot
simply convert the charge and then set the optimum gain value
(i.e. multiplier) --- that would simply use up the charge in
first place --- you have to use a preset gain value, which is
represented by the ISO setting.

Or, alternatively, you go get a D/A converter that does 17 or
18 bits with acceptable noise even at 80.000 as well as on 80
or less electrons, forget all about ISO and simply multiply
digital numnbers just right after the fact.  (Unfortunately,
no such beast seems to exist for digital cameras, especially
not for non-billionaire buyers.)

[1] Not all photons get converted: not only is no real-world
   sensor and no photon-electron converter perfect, but there
   are also unreceptive areas on the sensor.  Microlenses,
   while not without some potential problems themselves, can
   help ofset these areas, though.

[2] Possibly passing the charge through a gain multiplier for that
   specific cell (CMOS) or row or column of cells (maybe CMOS,
   certainly CCD) , to make all cells similar enough in their
   light response.

Not all sensors ave Charge Coupled Devices.  Canon's newer
DSLRs use Active Pixel Sensors of Complementary Metal Oxide
Semiconductors --- a completely different beast.  (CCDs are passive
and move charges to a possibly single read-out circuit, AP-Sensors
have an amplifier circuit for each pixel, just as one example.)

Aeh, no.  The noise is already there.

For example you don't always get the identical number of photons
due to the chaotic and random nature of the universe; just like
rolling dice: if you roll a D6 1000 times you'll get a sum near
3500, but if you once ... it could be 1, it could be 6.  So even
neighbouring pixels may get different intensities; and the less
photons, the worse (even though absolute differences are less).

Additionally, the electronics also introduce noise, however,
that is not bound so much to the charge size, but more often to
the temperature.

But with ah high ISO setting --- remember you amplify not only
the (weak) signal, but the noise as well, and the signal-to-noise
ratio is worse --- and a multiplication cannot improve the ratio.

But if you don't up the gain, you'll get very dark images.
(and no, 'pushing' after the A/D conversion doesn't buy you
anything, to the contrary: the converter also adds noise, and
that hurts weak signals much worse than amplified ones.)

You can push and pull chemical film, basically you can
develop longer and shorter.  The trade-off is worse
performance.  But since shadows and highlights act
differently to pushing and pulling (the shadows mostly don't
care, I understand), you would expose for the shadows (the
darkest parts to havet structure in them) and then later on
develop your specific image so that the highlights are just
as bright as you want them.  

That way you could force a low- or high-contrast scene to have
the contrast the film allows, and neither more or less.  (Of
course, this doesn't work for too much pulling or pushing.)

With digital equipment, you shouldn't try that: highlights
burn out unrecoverably *on exposure*.

You could try to expose so that the highlights were still
controlled and later push the darker parts of the image to show
texture.  (of course, you are limited by what your converter can
do for you, and by noise, again, and, especially when dealing
with JPEG as source materiak, you can quickly destroy graduated
changes into abrupt ones.)  Or you could try the 'contrast'
setting on your camera and see if that helps.

There is only one zoom lens for Nikon cameras?  Are you sure?

-Wolfgang

Why, sure, if you like bad pictures with dark shadows. Then who cares
about the noise?

Rule #1 in photography: If the light is bad, don't take the picture.

and rule #2 is like unto it: If you took the picture anyway, don't show
it to anybody.

Every photo I have ever seen where somebody insisted that he needed
more than ISO 400 was, in a word, terrible. Okay, there are some
exceptions with grainy ISO 1000 black and white film -- they are only
half terrible.