I "learned" method 1 (it's... obvious?), though my paper usually looked more like method 2 b/c I'd make marks to hold place. Otherwise my shitawful handwriting could have me interchanging digits all over the place.
Also, method 2 is more mathematically consistent. If I want to divide A into B, where B is written a*10000 + b*1000 + c*100 + d*10 + e*1, then long division has you doing:
1) find alpha (btwn 0 and 9) such that A*alpha*10000 <= B <= A*(alpha+1)*10000 (0 in this case)
2) find beta (btwn 0 and 9) such that A*beta*1000 <= B - A*alpha*10000 <= A*(beta+1)*1000 (9 in this case)
etc
The whole point is that 9*6000 = 54000. That's the closest you can get with only units of 1000; then you remove 54000 from the dividend and re-ask the same question w/the remainder. Method 1 just drops the 0s b/c this is obvious & doesn't need to be repeated.
I totally disagree about polynomial division. Method 2 is consistent with polynomial division. Say I'm dividing (x+1) into (x^2 -9x - 10) (example taken from first google hit).
Step 1: how close can I get using only units of x^2? Well (x+1)*x^2 = x^3 is not in the dividend, so 0*x^2.
Step 2: how close can I get using only units of x? (x+1)*x = x^2 + x. I can eliminate the x^2 term, leaving me with -10x - 10
step 3: how close can I get to the remainder using only units of 1?
etc (well we're done but you get the point)
The numbers long division asks the same questions: how close can I get using units of 1000? Units of 100? Units of 10? 1? It's basically the same problem but with a different "basis": in one case, 10^n, 10^(n-1),..,10,1; in the other, x^n, x^(n-1), ..., x, 1. You are NOT asking (as with method 1) "how close can I get to 56 using units of 1?" That's a simplification that helps you write down less stuff.
edit2: Oops, in addition to a different "basis," the polynomial division case doesn't have the restriction that the coefficients lie between 0 and 9. If you think about it more generally, it's still the same deal. Because 1*10, 2*10, ..., 9*10 cannot be expressed using (integral) units of 100. Similarly a*x cannot be expressed using (no coefs of 1/x or x allowed) units of x^2.
edit: AND since so many people here seem to think being college educated lends credence to their opinions, let me weigh in: i was a double major in aerospace & math. i'm getting my phd in computational fluid mechanics. I've had more math classes than most people in this thread... and that makes me an expert in division...!
That said, I've definitely run into teachers trying to tell me that my way of doing X is suboptimal/wrong/whatever and I'll lose points if I don't do it the way they were teaching. I can't remember how we were taught to find GCD & LCM in school, but the method my mom taught me was way faster. Or when we first learned systems of equations in middle school. I realized you could multiply an equation by something & add it to the other equation(s)... b/c of the equals sign. But I'll tell you I failed some assignments b/c I didn't do the substitution method we were being taught.
edit3: who cares that the numbers are large or small? If 54000 is scary, then the issue that needs addressing is... why is 54000 scarier than 54? They're just numbers. And again, the numbers are the same size in method 1, because method 1 is a simplification of method 2.