Implied Multiplication 3: You Can’t Prove It

This is the last of a series on our discussions, since I closed comments at the end of 2021, of Implied Multiplication First (IMF), the idea that multiplications written by juxtaposition, rather than with a symbol, are to be done before other multiplications or divisions. Last time, we saw that there is no “official” answer. Here, we’ll look at attempts to prove one view or the other. Most of them are variations on the misuse of the distributive property that was discussed in the original post; but there are some new things to learn from them.

For Strict PEMDAS: “You just need to add brackets”

Some people argue against IMF, and in favor of “strict PEMDAS”, according to which all multiplications and divisions are done as written, from left to right.

Samuel, from Tennessee, argued in July 2022 that, since we can indicate that a multiplication is to be done first by putting parentheses around it, there is no need for the IMF rule:

The answer given ignores specific other operators that are made specifically for those instances. Namely the bracket or braces.  Specifically if you’re wanting a multiplier (implied or otherwise) to apply first before other operators that is why the bracket and braces were specifically made.  The confusion of the issue is most schools skip the brackets and braces as unless you going into high level math operations they are never used.  As such most people never remember them or learn them.  Pemdas is just mnemonic device not the be all end all of the rule.  And adding a new rule instead of taking advantage of the already written rules breaks convention that convention has already accounted for.

Specific examples:

6 / 2(1+2) = X

In that equation X should equal 9. If however you want x = 1 you would simply write it

6 / [2(1+2)] = X

Anyways hope that helps clear the confusion.

It’s unclear what he thinks students are not taught; perhaps he means that only round parentheses, rather than square brackets and curly braces, are used in most math classes, and he feels that those different forms help here. He’s right that using extra parentheses lets us say what we want (regardless of which rule you go by); but which kind you use doesn’t really matter.

I replied, referring to the post he had read,

It’s certainly true that adding parentheses to clarify what you want to write is a solution to the problem from the writer’s perspective; alternatively, one can avoid the horizontal format for division and always write it as a fraction, as we recommend. But that doesn’t remove the problem for a reader who comes across such an expression, because he can’t be sure the writer sees things as you do. In that case, the main thing we need is to be aware that not everyone means it the same way, so that we need to ask the writer what rules they were using.

I also showed in the following post why the higher precedence of implicit multiplication might be considered correct. (But I still said we should avoid depending on it, because both readers and writers will always vary in their interpretations, even if rules were uniformly taught.)

He went on to argue that it isn’t “subjective”, a discussion I showed in the first of this series. In the end, he understood that the rules are not as well established as he had assumed (see the second of the series).

For IMF: “The factor belongs to the parentheses”

More people seem to write to us defending IMF, often not having noticed that I (tentatively) supported it.

An anonymous reader from Romania, in September 2022, started with a seemingly innocuous question, and eventually revealed that he was arguing for IMF. He initially wrote “4 ÷ x^2 = 4 ÷ x*x”, which is false unless you do all multiplications before division (as we’ve seen that some do). Then he wrote,

But if it is written in the form 4 ÷ x(x) with the first x being the common factor removed from x^2?

Is there a rule in mathematics that requires parentheses to be inserted before the common factor?

I replied,

I’m not sure what you mean.

The expression 4 ÷ x² means that you are dividing 4 by the square of x. There is no “common factor” to be “removed”.

Now, if you follow the (non-universal) “implicit multiplication first” rule, as explained in

Order of Operations: Implicit Multiplication?

then 4 ÷ x(x) would mean the same thing as 4 ÷ x², because you would be doing the multiplication first, though this would be quite an unnatural thing to write, and risky because of the ambiguity. This would be different from what you originally wrote, with an explicit multiplication: 4 ÷ x^2 = 4 ÷ x*x.

One would never do this “simplification”; but he evidently thinks that it should not be necessary to write 4 ÷ [x(x)]. He answered,

I don’t remember there being any rule that tells me that if I take out the common factor before division, I must put the square bracket. That’s why consider that if there is no multiplication sign between the factored number and the parenthesis, it means that the number belongs to that parenthesis, that’s how we say it, it belongs. So the square bracket in front of the factor is implicit and I do those operations first. I don’t want to create a controversy, I just want to understand the logic, because I can give my children the wrong advice by telling them that it would be interesting to impose a rule that no matter what type of equation you have, when you take out the common factor you put before the parenthesis factor.

I responded,

Again, you are using the term “common factor” in a way that doesn’t fit its meaning; I don’t think the term means what you think it means! A common factor of two numbers (or expressions) means a factor they both have in common; for example, 6 and 8 share the factor 2, and x and x² share the factor x, so 2 is a common factor of 6 and 8, and x is a common factor of x and x². None of what you write includes two different things with a common factor. So I’m struggling to understand what you are saying. Probably you just mean “factor”; you are “factoring out” an x from x^2.

A general rule I recommend is to put parentheses around anything that you change, just in case it might change the meaning. That may be more or less what you are saying. For example, when you replace a variable with a number, it is wise to put it in parentheses: When you replace x in “2x – 3” with -1, you need parentheses, “2(-1) – 3″ so that you are still multiplying by 2. If you just wrote “2 -1 – 3”, it would not have the same meaning. There are many other examples.

In the case you are asking about, writing “4 ÷ x^2 = 4 ÷ (x*x)” is appropriate, because it ensures that x*x is taken as a unit. In addition, this is part of the reason we usually use the fraction bar rather than the division sign you are using; it is clear that \(\displaystyle\frac{4}{x^2}=\frac{4}{x\cdot x}\). No one could misunderstand that!

I do think you will benefit from reading the page I referred to, which is, in part, about when you need parentheses to make a meaning clear. The point there is that some people do teach that “if there is no multiplication sign between the factored number and the parenthesis, it means that the number belongs to that parenthesis”, while others do not, so that using parentheses prevents misunderstanding. Furthermore, in your example, this prevents the error of taking it as you do in the first equality, but then changing to the explicit multiplication in the second, which changes the order. This doesn’t create a controversy, but works around an existing one.

We’ll see more about this idea of protective grouping as we proceed. Even when parentheses are not necessary (by whatever rules you are using), they can keep you from silly mistakes.

For IMF: “Absurd results for common factors”

This May, David from Australia argued for IMF, both catastrophizing and giving a faulty reason much like that last one:

In a discussion on this meme a posted article from you suggests that there is no good reason for completing the bracketed expression 8(14-8) before applying the division

I put it to you that there is a very good reason for doing so. If we don’t, mathematics as we know it collapses.

Consider this: 48 / (14*8 – 8*8) = 1.

Extracting the common factor, we get the equation 48 / 8(14 – 8) = 1

if you do the division first you get the absurd result that 48 / 8(14 – 6) = 36

To get the correct answer i.e. 1, you need to accept that implicit multiplication has a higher precedence than Multiplication or Division.

Would you like to comment and perhaps clarify the online article?

He had only read a snippet from Implied Multiplication, and not the following post, Historical Caveats, where I give arguments in favor of IMF. (Be careful about correcting someone when you don’t know what they’ve actually said.)

I responded, first discussing the proper way to manipulate expressions (which we’ll see later in an improved form), and eventually got to this:

In the same way, when you factor the denominator in 48 / (14*8 – 8*8), you need to put the entire factored denominator in parentheses, as 48 / (8(14 – 8)), if you want to be sure to preserve the meaning.

So if mathematics as we know it has collapsed, it is not the order of operations rule that did it, but your own carelessness with parentheses. You must always think about meaning before you rewrite an expression, rather than deriving meaning from an unthinking rewrite.

As I said in the post, “you unknowingly apply an alternative interpretation when you think you are just applying the distributive property.” By not adding parentheses, you are implicitly assuming that the multiplication will be done before the division (so that the parentheses have no effect, and can be omitted), which is what you are claiming to prove.

I hope that clarified things. The order of operations is a choice we make, not something we can prove to be mathematically necessary. As I said in Order of Operations: Historical Caveats, there are some good reasons (human more than mathematical) for preferring the implicit-multiplication-first approach; but what we actually do has to be agreed upon by mathematicians, rather than proved, and that agreement is not currently universal (at least in the U.S.).

One “human” reason for preferring IMF is that it is easy to accidentally do what he does! But it is better to consciously express the correct meaning.

For IMF: “Substitution requires it”

Pete from Georgia, like David, in July argued that substitution of a subexpression implies IMF:

Your article piqued my interest and I read it in its entirety. It explained the dichotomy of solutions very well. What caught my attention was your reference to what you call “the sticky parentheses.” I would have preferred you call it “the sticky GCF or coefficient” though both appellations nicely describe how I view the problem.

MY QUESTION IS THIS: IS NOT THE 2 OUTSIDE THE PARENS THE GCF/COEFFICIENT OF THE GROUP AND AS SUCH IS IT NOT “TIED” TO THE EXPRESSION?

Here’s my reasoning: Let us take two expressions: (2*1+2*2) and 2(1+2). By themselves, they will both compute to 1 and therefore they are equivalent values in either form. Factoring a binomial does not change its value. If we were to substitute the first one as in 6÷(2*1+2*2) it will solve to 1 using PEMDAS and the order of operations. And of course, if we were to simplify that expression by factoring out the 2 it will of course be the YouTube form of 6÷2(1+2). It is problematic (for me) that when two expressions of equivalent values when taken alone seem to change their value when put into the formula with other operands! In order to arrive at an answer of 9 the 2 would have to be ripped away from the remaining factor of (1+2). I guess it wasn’t sticky enough!

I replied:

Did you miss the fact that the whole “sticky” idea is an error? My point there is that parentheses are “sticky” only on the inside (that is, they hold together what is inside them, so that it has to be evaluated as a single entity), and the “P” rule says nothing about how they relate to whatever is on the outside. That is covered by the other rules; and the whole issue is whether there is a rule that implied multiplication is to be done before division. All you are doing is relating the idea to factoring, which is an application of distribution; it is still wrong. Whether you relate it to the parentheses or to the factor outside them, the a(b+c) construction does not “stick together” merely because of distribution/factoring.

He later referred to a description of “equality-preserving transformations” of algebraic expressions, which says,

The simplest equality-preserving transformations involve arithmetic. For example, you know that (2)(14) = 28. Transforming expression (2)(14) into 28 is an equality-preserving transformation.

You also know that x(y+z) = xy + xz. That means

(5w) + (x(y+z)) = (5w) + (xy + xz)

since you can perform an equality-preserving transformation on a subexpression.

Applying this to 6÷2(1+2), he said,

This proof manifests itself in this way: By substituting a sub-expression with an equal sub-expression. So if we were to replace the 2(1+2) with the equal (2*1+2*2) it would have to resolve to 1 using PEMDAS.

So he is asserting that we can directly replace \(2(1+2)\) in \(6\div2(1+2)\) with \(2\cdot1+2\cdot2\) to get \(6\div(2\cdot1+2\cdot2)=6\div(2+4)=6\div6=1\).

Preemptive parentheses

I responded:

No reasoning can prove that it must be interpreted one way or another; as I have said over and over, this is not a matter of mathematics, but of linguistic convention.

Your reference to “equality-preserving transformations” actually agrees with what I said, and not with your claim. You evidently missed the details of their example:

You also know that x(y+z) = xy + xz. That means

(5w) + (x(y+z)) = (5w) + (xy + xz)

since you can perform an equality-preserving transformation on a subexpression.

Note the parentheses he put around the sub-expression in both expressions, just as I said you need to do. He could, in this case, have just said

5w + x(y+z) = 5w + xy + xz

because the parentheses in this case don’t change anything (because of a combination of order of operations and the associative property); but he carefully put parentheses around each term on the LHS before doing the substitution in order to demonstrate the level of caution that is needed in doing a substitution. You have to at least think those parentheses, and omit them only when you know they are not needed. This same practice would prevent all the errors I demonstrated in making mechanical substitutions — and would also prevent your mistake.

I had previously quoted what I’d said to David about substitution; now I restated that with the preemptive parentheses:

Your thinking is equivalent to what I discussed in the section headed “Misapplying the distributive property”. You are just factoring (undistributing) rather than distributing as that person was doing.

In mechanically replacing (2*1+2*2) with 2(1+2), you are ignoring the order of operations, or rather, assuming that a literal substitution never affects the order. Here are some other examples in which such a substitution is inappropriate:

Given 2x, if I mechanically replace x with y+3, I get 2y+3, which is wrong; I have to see it as 2(x) and then replace (x) with (y+3), using parentheses, to get 2(y+3) in order to get an expression that means 2 times y+3. Without the parentheses, we are only multiplying the y by 2, and not also the 3.

Given 3², if I mechanically replace 3 with 1+2, I get 1+2², which is wrong; the original expression has the value 9, but the new one is only 5, because we are now using only the 2 as a base. I need, again, to use parentheses to preserve the intended order: (3)² = (1+2)² = 9.

Given (2*1+2*2)², if you factor 2 out of the base and write 2(1+2)², you will be wrong; the original expression has the value 6² = 36, while the new expression has the value 2(3²) = 18. In order to make an equivalent expression, you need to write (2*1+2*2)² = (2(1+2))², replacing a parenthesized expression with an equivalent parenthesized expression, so that the multiplication will be done before the exponentiation.

In the same way, when you factor the denominator in 6÷(2*1+2*2), you need to put the entire factored denominator in parentheses, as 6÷(2*1+2*2) = 6÷(2(1+2)), if you want to be sure to preserve the meaning.

I added,

Comparing what you say with your reference, you would need to replace, not just 2(1+2), but (2(1+2)), with (2*1+2*2), in order to be sure not to change the order of operations.

But in our expression 6÷2(1+2), if strict PEMDAS is followed, then 2(1+2) is not properly a subexpression, as it is not evaluated as a unit. (And we aren’t changing its value to 3.) In the same way, I wouldn’t call “2 + 3” a subexpression of “1*2+3*4”, and expect to be able to replace it with 5. Why? Because I couldn’t write the entire expression as “1*(2+3)*4″.

So the problem with your attempted proof is that you are implicitly assuming one interpretation of the order of operations in order to prove that only that order makes sense. This is circular reasoning.

So the bottom line is as I have always said: Rather than try to prove one interpretation “mathematically correct”, we need to recognize that all three rules make some sense, and have been taught, so that caution is the only rule.