Which direction does current flow?

[ReverseEMF]

Electrons don't really flow. They just jump to the next atom, which displaces another electron, etc., etc. The holes flow, but then holes aren't really anything.

I think you're confusing current flow in a conductor with current flow in a semiconductor.

• Conductor: Movement of free electrons [a property of metals].
• Semiconductor: More like what you're describing [a property of specially doped semimetals or more exotic alloys].

[ReverseEMF]

And, to more properly answer the actual OP question: Actual physical direction of current flow depends on the "charge carrier" involved:

• Electrons, being negatively charged, are repelled by negative potential.  Thus, they will flow from negative to positive, or from more negative to less negative.  This is the most common case in electronics, as metals, which by nature, possess an ambient cloud of "free electrons", are the most prominent conductors.  Even in a Semiconductor, the charge carrier is alway electronics--even in a P-type semiconductor.  Holes are just a concept [conceptual charge carriers], and are actually produced by the movement of electrons.
• Protons, being positively charged, are repelled by a positive potential, and thus move in the exact opposite way, electrons do.  The closest case, in electronics, that I can think of, where protons serve as the charge carriers, is in a smoke detector that employs Americium--which gives off alpha particles.  And, alpha particles are the nuclei of Helium [which contain protons].  
• Ions are considered charge carriers, but even their charge is the result of a summation of the charges on protons and electrons [and, in fact, my smoke detector example is really an example of ions as charge carriers].  Ions can have an overall charge that is positive or negative.  Their direction of movement is determined by this aggregate charge.  If positive, then they move towards a negative potential [or less positive potential].  If negative, vice versa.
• Holes, as mentioned above, are conceptual charge carriers.  They don't exist as a physical entity, but they are listed as a type of charge carrier. Holes always have a positive "charge", and thus, move like a proton.

[pwillard]

We can blame Benjamin Franklin all day long for getting it wrong... but in the end... its just easier to use the "conventional" flow concept (Hole Flow, if you must) as it matches the direction that a DIODE in a schematic diagram will point.

Don't make me use diodes backward...   ;-P

[wvmarle]

Ions [/i]are considered charge carriers, but even their charge is the result of a summation of the charges on protons and electrons [and, in fact, my smoke detector example is really an example of ions as charge carriers].

Another example of ions as charge carriers is in electrolyses, and in general the conductance of water.

Don't make me use diodes backward...   ;-P

Not just diodes... transistors as well.

[ted]

This

and this

[pwillard]

So the Answer based on TED's empirical evidence...

"Which way does current flow?"    -->  Whichever way it *can*.

[ted]

Bingo

[MarkT]

Holes, as mentioned above, are conceptual charge carriers.  They don't exist as a physical entity

You'd need to define "physical entity" fairly carefully to make that claim watertight, and probably
lose the baby with the bathwater in the process.  To a semiconductor physicist holes are very real...

[Grumpy_Mike]

To a semiconductor physicist holes are very real...

The are, they even have mass, which means the electrons in certain circumstances, have negitave mass. That means they are repelled by a gravitational field.

So go and patent your anti gravity machine now.

[ReverseEMF]

You'd need to define "physical entity" fairly carefully to make that claim watertight, and probably
lose the baby with the bathwater in the process.  To a semiconductor physicist holes are very real...

Yes, holes are real, in the sense that a hole in the ground is real.  If I try to set a flower pot on a "hole" it will not be supported, and I can fill it with water.  It's an absence of dirt defined by the dirt around it.  But, if I dig a hole next to it, and toss the dirt I dig up, into the previous hole, the hole will appear to have moved.  But did it?

I can conceive of a model where holes in the ground "move", and it may work for certain physical systems -- like, maybe, some sort of "outback whack-a-mole", or stop motion animation where the hole becomes a character.  But, the model falls apart if I actually try, on hands and knees, to shove the hole around the ground it's dug in.

If I try to actually move a hole in a crystal lattice--the lattice will shatter.  The hole doesn't move, but it's useful to conceive of a model of it moving.  But, that's all it is -- a model!  And any physicist is deluding him/herself if they believe holes actually move.  Believing models are real is a trap that even scientists fall into, and a trap that limits their thinking in a way that can hinder innovation.

And to make this crystal clear [pun?], I'm NOT debunking holes, I'm pointing out that the idea that holes actually move is a delusion, in the same way, anthropomorphising an android is delusional.

[Southpark]

You'd need to define "physical entity" fairly carefully to make that claim watertight, and probably
lose the baby with the bathwater in the process.  To a semiconductor physicist holes are very real...

Agreed. All these 'physical' things we see today appear to be formed by some forces (or whatever they call them which scientists probably don't know the origins of, or how they exist etc). Those forces (or something) somehow form (or give us the impression of) things/entities/objects that people are able to define through observation, or theory, or measurements etc...... what appears to be physical definable things, but are a result of 'forces' (or something, which themselves aren't understood in terms of what they are, where and why.) Eg.... these forces somehow combine to give impressions of 'electrons', 'holes', 'rocks', 'wood', 'atom', chemical 'molecules', electric/magnetic fields etc. So at the base of it.... if OP is talking about current, or charge flow..... it's sort of like talking about how the forces (probably in some matrix pattern across our universe) bans together to give the impression of an 'object' or some lumped things moving.

So...... as long as we define something clearly, and as long as we agree upon following particular definitions, then this helps to avoid issues with the communication.

[ReverseEMF]

Agreed. All these 'physical' things we see today appear to be formed by some forces (or whatever they call them which scientists probably don't know the origins of, or they exist etc). Those forces ...<snip>

I wasn't saying holes are not real -- just that they don't really move.  But, I see my error. : I got my wording wrong.

[Grumpy_Mike]

I wasn't saying holes are not real -- just that they don't really move.

Of course they move, how else can they have a mean free path?

[ReverseEMF]

Of course they move, how else can they have a mean free path?

What, exactly, is moving?

[Grumpy_Mike]

What, exactly, is moving?

The hole.