1.  No, I was thinking that you'd previously proved "If A and B, then C" and were proving "If C and D, then E" and were going to conclude "If A, B, and D, then E" but you have no way of knowing if the first proof is really valid just because you remember it being valid. 

2.  Incorrect.  You don't KNOW which way is up, but there is a direction that you THINK is up.  The indication is not guaranteed but it is there. 

/facepalm "not necessarily" is the reason it's only likely.  [edit:  Never mind what I deleted.][edit2:  Also, please remember that Occam's Razor applies to of the available explanations and considering the available data.  I suspect that the source of your problem may be in there somewhere.]

@ donathos: "ETA"?

1. Wait, "If A, B, and D, then E"... How would that even work?

2. There is no direction you think is up; you're totally dizzy, it's completely foggy, you have absolutely no indication which way is up.  You have to take a wild guess.  Just as it is, I'm saying, to take empiricism.

Occam's razor says the least complex explanation must be the most likely, and that's what I meant was not necessarily true.  And no, you could not say it's likely that the least complex is likely, either; nothing indicates that possibility.

1.  Simple, or at least it's simple for me.  Logic class was way too easy for me. 
If you have
"If A and B, then C" (call this P1)
and you have
"If C and D, then E" (call this P2)
and you presume A, B, and D, then:
Since A and B are true, then you use P1 to get C.
Now you have A, B, C, and D. 
Since C and D are true, then you use P2 to get E. 
Voila!

2.  But that is the equivalent of saying that you have no sense data at all, which is not what we are discussing.  We have no 100% trustable sense data. 

To include my second edit:  "Please remember that Occam's Razor applies to of the available explanations and considering the available data.  I suspect that the source of your problem with it may be in there somewhere."

1.  Oh, so you're not giving a 3-premise argument (that's what I was confused by); your simply restating a conclusion/premise as it's two premesis.  In any case, it's still a question of if your premesis are true, so forgetting what came before originally wouldn't affect the logic, unless your restating them all at the same time, and in that case, you wouldn't have to rely on memory.

2.  We have no data that is even 1% trustable; that's what I'm saying.  Data can support anything you want it to, because it doesn't speak for itself (as we agreed).

I'm not quite sure what your edit does to the situations.

1.  But see, the "If A, B, and D, then E" argument hinges on the previously proven "If A and B, then C".  So if however that was proven is erroneous, then it all falls apart.  I'm presuming that there's more work behind it than sheer assumption; that the simple thing stated here is not the whole story.  And in any case who said short term memory was safe? 

2.  Okay fine, it's 0% trustable.  But the evidence still exists.  As opposed to having absolutely no information, regardless of quality, available. 

3.  Your objection to Occam's Razor was "it doesn't follow that the idea that assumes the least amount of statements is the most likey correct, because it is unknown if reality, in reality (lol), depends on only a few things or a trillion things (or infinite things)."  It seemed to me that you were saying 'but there could be completely unknown things that are unaccounted for that could make the Occam's Razor choice a worse one'.  And that is an inapplicable objection because OR concerns itself only with choosing considering the available data.  And other, more complete explanations may exist that account for this hypothetical data but again OR is only for choosing among the available explanations.  At the time of the choice, OR is a guideline to the best explanation but is not guaranteed to be right. 

Does the expanded version help any?

Edited To Add (that section wasn't in my original post).

Oh now here's a rarity...someone else who actually understands Occam's Razor.  Most people tend to miss the "all things being equal" bit =P  And of course once you've missed that it is pretty easy to misstake the phrase "tends to be" to mean "absolutely must be".

Even that is somewhat of a lamen's definition though, we can blame Matthew McConaughey for that I suppose.

Here's my view of Occam's Razor:  (and this'll be fairly abstract; I'm aiming for clarity)

There are three data points that need explaining, A, B, and C.

Three different theories are proposed to explain them.  Theory 1 explains A & B, but does not explain C.  Theory 2 explains A, B and C, and also requires that we accept the yet unknown/unproven data point D.  Theory 3 explains A, B and C, and does not require data point D to be true.

I believe that Occam's Razor says that, based on these conditions and these alone (i.e. "all else being equal"), we give tentative agreement to Theory 3.

Theory 2 is complicated beyond necessity; data point D is clearly not needed to explain A, B & C, as Theory 3 demonstrates.  And so there is no need to invent data point D.

However, and speaking to my corollary, Theory 1 is "too simple."  It doesn't do what we need it to do, which is to explain all three of our given data points, and it too must be rejected in favor of Theory 3.

Now, like I've said, this is an epistemological exercise, not a metaphysical one: it may be the case that Theory 3 is false, and that data point D exists and Theory 2 is true.  It's just that, unless we have data point D, there is no call to take Theory 2 above Theory 3; we fit our working theory to the available evidence, no more and no less.

And so, let me try to directly answer your question as phrased:

Not the most likely, exactly.  Just that the most simple solution is the only one we're justified in adopting.

Gar, I hate sounding "academic."  Does that make a lick of sense to anyone other than me? :)

But I do not see why the one that best explains the situation (without explaining enough or something else additionally) means we can only take that one as the one to assume.