Sunday Morning Greek Blog

If you like this article, you may also like 8 ÷ 2(2 + 2) = 1: Why PEMDAS Alone Is Not Enough or 8 ÷ 2(2 + 2) = 1, Part 2: A Defense of the Linguistic Argument.

I’ve gotten caught up in the PEMDAS/Order of Operations (Parentheses/Brackets, etc.; Exponents; Multiplication & Division [left-to-right]; Addition & Subtraction [left-to-right]) discussions on social media. I must admit it is frustrating to see how religiously some people cling to an imperfect and almost thoughtless understanding of PEMDAS. It is a convention that, as currently practiced, has no theoretical soundness and is clearly NOT intuitive. In linguistic terms, I would describe the current application of PEMDAS as passive, that is, it doesn’t seek to recognize certain theoretical and intuitive constructs in mathematics, and even undermines them, and as such represents a primitive view of mathematics.

By “primitive,” I mean that some of the proponents of a strict PEMDAS are childlike in its defense, resorting to emotion and circular arguments based on their assumptions. No one has offered any proof that what they believe about its inner workings is in fact how people have historically interpreted some of these expressions. Nor have they offered anything to counter the various arguments and peer-reviewed documentation I have offered except by returning to their assumption and repeatedly showing us how they think the target expression should be worked even as we question their premise. PEMDAS pays too much attention to the individual signs and symbols, but it lacks the sophistication to understand the “syntax” or intention of certain conventions of expression we have developed over the centuries.

PEMDAS as practiced today is like reading the synoptic gospels of the New Testament without reading the gospel of John, or like reading the New Testament without also reading the Old Testament to understand its foundations. If you’re not biblically minded, it’s like reading Shakespeare without a 16^th century dictionary at hand. I aim to fill that void here and make PEMDAS something so theoretically reliable that we won’t need to have wasteful discussions in social media about who’s right and who needs to go back to elementary school. My purpose is to propose such a theoretical structure whereby we can agree on one rule to address the issue of these viral problems.

Properties

I want to begin with a discussion about the basic mathematical properties we all had to learn at the beginning of the semester in Algebra. Properties are demonstrable principles in mathematics that will help lay the foundation for what some have begun categorizing as “grouping.” But more on that later.

Associative and commutative properties are intuitive. By that I mean that it’s obvious to most that if we have the expression

a * b * c

it does not matter how we “group” the terms, that is, how we “associate” them within parentheses or how we “commute” terms in the order of presentation, we get the same result. Every permutation of the variables multiplied together yields the same result. They’re all cofactors of the product, after all, so you get the same result every time. As such the following expressions are ALL literal identities of each other.

(a * b) * c ≡ (a * b)c ≡ (ab)c ≡ a * (b * c) ≡ a(b * c) ≡ a(bc) ≡ c * (a * b) ≡ c(a * b) ≡ c(ab) ≡ b * (c * a) ≡ b(c * a) ≡ b(ca), etc.

Distributive property is intuitive as well because it’s the undoing of factoring:

a(b+c) ≡ (ab+ac)

That expression represents implied (or implicit) multiplication, and the result demonstrates that the distributive property is a form of grouping equivalent to putting parentheses or brackets around the entire expression. As a type of grouping, then, it belongs to the P in PEMDAS. In addition, I would argue that even when there is an explicit multiplication sign used for the distributive property:

a * (b+c) ≡ (ab + ac)

this does NOT negate or override the application of the distributive property, and the expression thus written is also considered grouped and should be treated as if it has parentheses or brackets around it. Why? Because we know that if two expressions are equal or identical to a third expression, the other two expressions are identical as well:

a(b+c) ≡ a * (b+c).

Since both expressions are identical, it is necessary to consider both as grouped, as implied multiplication by the action of the distributive property at least, and in the case of the first one, by the syntax of implied multiplication. As such, my first proposal to strengthen PEMDAS and place it on a sounder theoretical ground is this:

Proposition 1: Mathematical properties should be considered at the same level of grouping as parentheses in PEMDAS, and when so recognized, the human person working such expressions has both the responsibility and authority, by virtue of any applicable mathematical properties, to place such expressions within the appropriate form of brackets to indicate priority given to the calculation. In doing so, the human person working the expression should ensure that any units of measurement or counting associated with the expression are not affected by so organizing.

The Trouble With Computers and Calculators. Most calculators do not recognize grouping characterized by this implicit syntax (I know syntax is primarily a linguistic concept, but for math, it refers to the arrangement of the signs and symbols, including the numbers, of an expression and how those influence the interpretation of the expression). You typically must apply the parentheses or brackets around the expression to force it to do so. Without parentheses, calculators assume all else in the syntax of an expression is explicit.

Implicit Multiplication Generally

Having trouble accepting implied multiplication is a form of grouping? What is a number with an exponent if not implied multiplication? The term 3³, or in calculator language, 3^3 is 3 * 3 * 3, but we don’t undo the exponent notation if a division sign comes before it. Why? Because we’ve given this form of implied multiplication priority over signed multiplication in PEMDAS. The implied multiplication is grouped into the simple expression, algebraically speaking, of aⁿ. If you do deconstruct the implicit multiplication of an exponent, you’ve obviously taken PEMDAS to an unacceptable extreme.

PEMDAS does not address what to do with a factorial expression.

4! = 4 * 3 * 2 * 1

Algebraically, this is expressed as

n! = _nP_r =n(n – 1)(n – 2)…(n – r + 1])

where _nP_r is the number of permutations (P) of n things taken r at a time and n = r.

The factorial is also considered a form of implied multiplication, then. However, since PEMDAS doesn’t explicitly deal with factorials in the order of operations, how should we treat it? Algebraically, the first term in the grouping doesn’t have parentheses around it, so if we come across a ÷ n!, are we going to treat that as (a/n) * (n – 1)! then? Absolutely not! I’ve already hinted at why. It’s considered a grouping of factors implicitly multiplied. Because it’s a grouping, then, we would also treat it as if it were either an exponent or within parentheses. If these two types of implicit multiplication (powers and factorials) are generally treated at the E level in PEMDAS, why shouldn’t the implicitly multiplied expressions a(b + c) or even ab for that matter, or any variation thereof as described above, be treated at least at the E level, if not the P level of PEMDAS?

This leads me to my second proposition and a subsequent corollary to strengthen PEMDAS and place it on a more solid theoretical ground:

Proposition 2: In the absence of properties, any terms or expressions using implied multiplication should be considered at the grouping level of Parentheses in PEMDAS, and when so recognized, the human person working such expressions has both the responsibility and authority, by virtue of the implied multiplication, to place such expressions within the appropriate form of brackets to indicate priority given to the calculation. In doing so, the human person working the expression should ensure that any units of measurement or counting associated with the expression are not affected by so organizing. [Added 7/6/23] Within the grouping level, exponents and factorials should be processed first, then any implied multiplication would follow. So 3x^2 (without parentheses) should be evaluated as 3(x^2) and NOT (3x)^2.

Corollary 1: Any expression using explicit or implicit grouping markers (parentheses, brackets, juxtaposition, etc.) should NOT be rearranged in such a way as to undo the intention of the syntax or to fit certain elements into the framework of a property. For example, an expression in the form a/(bc) (explicit grouping) or a/bc (implicit grouping) should not be undone and reworked to become (a/b) * c or (ac)/b. NOTE: Some on social media tried to say the distributive property worked as follows:

b ÷ a(x+y+z) = [b * (x+y+z)]/a.

It’s ridiculous to think that a coefficient (or cofactor) should become a divisor in this situation.[1]

Let me also add here a call to return to standards. If the ISO standards don’t recognize the obelus anymore, then let’s just stop using it altogether. We can retain references to it for historical purposes, but in formal education (including textbooks), edit it out. People have tried saying the obelus is different from the vinculum, but again, I’ve never seen any formal explanation as to why this is, because the obelus isn’t a standard! I’m using it in this article because that’s how the expressions appear in the social media threads.

Excursus on Implicit Functions—Added 08/29/23 (from a post I made on Facebook), and lightly edited on 04/01/24.

Exponential terms don’t require a standard operator sign (+-x÷/) but rely on juxtaposition of the exponent.

Roots do not require a standard operator sign but the juxtaposition of the superscripted power and value in relation to the radical or of a fractional exponent. In some settings, the radical is NOT fully evaluated unless the answer can be expressed as a rational number. If the not-fully-evaluated radical is in the denominator of a fraction, the fraction must be rewritten so the radical is only in the numerator.

Factorials don’t require a standard operator sign but rely on the juxtaposition of a punctuation mark.

Cosine, sine, tangent, etc. don’t require a standard operator sign, but rely on the position of the measurement next to an abbreviated form of the word, and in the case of inverse trig functions, the juxtaposition of a -1 “exponent.”

Logarithms don’t require a standard operator sign but rely on the juxtaposition of a subscripted base (as needed) next to the abbreviated form of the word and the position of the value after the word/base combination.

Absolute value does not require a standard operator sign but juxtaposes, or brackets, the value between two pipes.

A fraction written with a vinculum does not require a standard operator sign that typically disappears when the operation is performed. The vinculum typically only disappears when the fraction can be represented as a whole number (mixed numbers, which consist of a juxtaposed whole number and fraction [implicit addition!!], typically have to be converted to an improper fraction when working them in an expression) or when, during the evaluation of the expression, the denominator divides evenly into the numerator. The fraction is a vertical juxtaposition of the two values with the vinculum indicating the grouping in the denominator and the relationship with the numerator. This is implicit division without actually dividing the two values to create a decimal. A fraction is considered a single value and NOT necessarily a division problem in the context of an expression. [This paragraph expanded from original on 04/01/24.]

Now while some of these functions do require parentheses to group the values being evaluated, NONE of them have a standard operator sign extant outside of any values in parentheses. These are all implicit functions, that is, they essentially have an implied set of brackets (or parentheses or some other grouping symbol) around them so they’re taken as a unit, or operand if you will. We don’t break out the implied multiplication of the exponent or factorial and supply the signs when evaluating the expression. Neither do we break out the implied addition when dividing by a mixed number, except in the manipulation of the mixed number to an improper fraction. In my opinion, that belongs in the P (parentheses) step of PEMDAS.

In the same way, an expression such as 2(2+2) is an implicit notation (no operational sign outside of the parentheses), so it belongs in the category of implicit functions. Operator signs break up an expression into operands. All this talk about “terms” that some have promulgated in the social media pages is meaningless, especially since the given expression above is technically one term by their own definition. We can debunk the contention 8 ÷ 2 should be treated as a fractional coefficient to what’s in parentheses. The obelus indicates we have two operands: 8 and 2(2+2). As an implicit function, with presumed parentheses around it, the 2(2+2) stands alone as the divisor in the expression. It should NOT be separated.

By the way, here’s how Wolfram defines “operand”: “A mathematical object upon which an operator acts. For example, in the expression 1×2, the multiplication operator acts upon the operands 1 and 2.” Implicit functions should have priority over explicitly signed operations and would need to have its own priority in PEMDAS for the functions included in that category.

End excursus.

Creating the Theoretical Foundation for PEMDAS.

Are you following me so far? Because I’m about to make it even more interesting. PEMDAS is neither intuitive nor theoretical (at least as applied) because there’s no proof to show why any one operation should be given preference over another. It is a convention that has rather haphazardly come into being historically without, as far as I can tell, any serious discussion about its theoretical or practical validity. As such, it does NOT mathematically or logically carry the same weight as properties.

The “intuitive” thing for a person to do just seeing a complex expression for the first time would be to simply work left-to-right through the symbols because we learn to read left-to-right in most languages. So such a person would treat the following as if it were in a series of brackets:

2 * 3 + 4 ÷ 5 – 6 = –4

becomes

{[(2 * 3) + 4] ÷ 5} – 6 = –4.

We used to play a game like that in 5th & 6th grade math class, where the teacher would give a string of numbers and operators (occasionally they’d put some operation in parentheses to see if we were paying attention) and we’d have to keep up with the math by taking the total as each new operator-number combination was announced and perform that operation on the running total. There was no going back through the string and figuring out PEMDAS or whether the distributive property applied. Whoever spit out the correct answer first got to advance toward the front of the class.

When we’re talking about PEMDAS, then, we’re talking about pedagogical construct that, historically speaking, seems to have been formulated rather hastily and loosely without much theoretical forethought. First, PEMDAS doesn’t account for everything that could be happening in an expression. I used the example of the factorial expression above. I’ve never seen anything in any explanation of PEMDAS that suggests where such an expression would fall. The factorial is important, for example, for calculating permutations. Treating a factorial in the wrong level of PEMDAS can potentially create false expectations about the probability of some event, like drawing winning lottery numbers. Exponents seems to be the most logical place because the factorial is a form of grouping by implied multiplication just like exponents are.

Second, in practice, PEMDAS is decidedly passive. It leaves too much room for interpretation, especially when historically, people have considered terms with implied multiplication to be grouped, bound, and inseparable. The PEMDAS absolutists in the social media threads on such problems have no respect for that historical reality. They suggest that the older generation’s understanding of PEMDAS is outdated when the real problem is their methodology and lack of critical thinking skills. This lack of critical thinking is becoming increasingly prevalent in American education today. Instead of a passive PEMDAS, we need an active, robust PEMDAS that focuses more on intentional grouping and a more sophisticated definition of grouping based on the assumptions made in the syntax of algebra.

Third, strict PEMDAS, at least as practiced by the absolutists, is no better than a calculator, because it fails to consider the implications of implicit multiplication. A calculator may be able to perform calculations quickly, but it doesn’t have a “mind” like you and I do to analyze the syntax and context of an expression.

The Priority of Operational Signs. Now that we have addressed the grouping issue and the weak nature of PEMDAS as currently practiced (if the absolutists are to be believed, anyway), it’s time to address the basic functions of mathematics: multiplication and addition. Since multiplication and addition are the primary operations mathematics, division and subtraction are respectively subordinate to them. We all can agree, I suppose, that division is multiplication by the inverse of the divisor, and subtraction is addition performed on a negative number. So while we’re beefing up the order of operations with a more active and robust use of parentheses to ensure any expression is solved the way its creator intended, let’s get rid of the obelus and the slash (solidus) as well and make everything about multiplication and addition. After all, isn’t that what we do when we divide by a fraction? We manipulate the expression (invert and multiply) to perform multiplication first, so wouldn’t that argue for the priority of multiplication in that instance? Let’s use only the vinculum to indicate “division,” and let’s put parentheses around a fraction with a vinculum so there’s no question that the vinculum is grouped. Fractions are, after all, more accurate than decimals if the decimal equivalent is not finite.

Let’s change all subtraction expressions to adding negative numbers expressions. Let’s keep all expressions with radicals as exponent expressions as much as possible. Let’s focus more on ensuring the proper ORDER of operations using parentheses and brackets rather than assuming everyone treats every expression in a uniform way. Obviously, that’s not happening right now. We can’t have a system of math where people are coming up with two (or more) answers to the same expression because they’ve been taught different things about it or work from a different set of assumptions. We need to firm up PEMDAS so we can have a uniform understanding of what fits where and restore uniformity. Without that, we’re wasting precious time arguing about ambiguities. It’s time someone cleared those ambiguities up. That’s why I’m stepping forward here to offer a theoretical solution.

One more question: If multiplication and addition are the primary operations of mathematics with division and subtraction respectively subservient to them, what is the theoretical reason then that multiplication and division are treated equally left-to-right? Why wouldn’t multiplication and addition be given priority over division and subtraction? I’ve never seen an explanation for this. After all, there are some out there who have a multiplication-first view of PEMDAS. I’m not sure I can fault that view given that multiplication has formal properties associated with it while division does not.

Conclusion

This whole process of “thinking out loud” in the social media posts has been, in my mind, anyway, a helpful discussion. In the process, I have more firmly codified my own beliefs about PEMDAS, as I’ve done here, and have continued to do so over the past year. I am even more firmly convinced that those who haven’t gotten the same answer as me on the more popular manifestations of this PEMDAS debate have a lack of theoretical understanding that should be corrected. I hope that I have filled that gap and that I have convinced at least some of you that the answer is 1, not the square of the numbers in parentheses (see Demonstration Proofs 1 & 2 below). I offer my proofs below of the three most popular expressions as a demonstration of application to my principles.

I do think PEMDAS is a valuable pedagogical tool, when properly understood with solid theoretical underpinnings. Professor Oliver Knill has repeatedly said the only way to ensure a problem is worked the way one means to is to overuse, if necessary, parentheses and other groupings.[2] Otherwise, ambiguous problems remain open to interpretation, and we’ll never get any kind of uniformity. My attempt here is to promote a uniformity of understanding so that we only get one answer for each unique expression.

Demonstration Proofs.

Proof #1

Given

36 ÷ 6(2 + 2 + 2)

Grouping by distributive property and implied multiplication

36 ÷ [6(2 + 2 + 2)]

Multiplication by inverse and grouping under vinculum

                    1

36 *     —————-   

            [6(2 + 2 + 2)]

Application of distributive property

                           1

36 *     ———————–

             [6(2) + 6(2) + 6(2)]

Multiply terms in brackets

                      1

36 *     —————-

           [12 + 12 + 12]

Add terms in brackets

             1

36 *     —-

            36

Cancel common factors and simplify

1

QED

Proof #2

Given

8 ÷ 2(2 + 2)

Grouping by distributive property and implied multiplication

8 ÷ [2(2 + 2)]

Multiplication by inverse and grouping under vinculum

                  1

8 *       ———–

            [2(2 + 2)]

Application of distributive property

                    1

8 *       —————-

            [2(2) + 2(2)]

Multiply within brackets

               1

8 *       ——

          [4+4]

Add terms in brackets

           1

8 *       —

           8

Cancel common factors and simplify

1

QED

Proof #3 (From Professor Oliver Knill’s “Ambiguous PEMDAS” Web page. When presented with this problem, 58 of his 60 calculus students got the following result indicated in the proof. The other two treated (3y – 1) as the expression under the vinculum. NONE got the answer promoted by the PEMDAS absolutists.)[3]

Given

2x/3y – 1; x = 9, y = 2

Group by implicit multiplication and vinculum

(2x)

——    – 1

(3y)

Substitute

[2(9)]

——-    – 1

[3(2)]

Multiply in brackets

18

—-    – 1

 6

Simplify

3 – 1

Subtract

2

---

[1] Professor Oliver Knill from Harvard (Ambiguous PEMDAS (harvard.edu)) cites Lennes, 1917. [No title given]: “When a mode of expression has become wide-spread, one may not change it at will. It is the business of the lexicographer and grammarian to record, not what he may think an expression should mean but what it is actually understood to mean by those who use it. The language of algebra contains certain idioms and in formulating the grammar of the language we must note them. For example that 9a² ÷ 3a is understood to mean 3a and not 3a³ is such an idiom. The matter is not logical but historical” (emphasis in original). In scans from Lennes printed article, these statements appear: “A series of operations involving multiplication and division alone shall be performed in the order in which they occur from left to right.” However, the subheading in the very next paragraph says: “The Above Rule Contrary to Actual Usage,” and the author goes on to say: “It would, however, follow from this rule for carrying out multiplications and divisions in order from left to right, that

9a² ÷ 3a = (9a² ÷ 3) × a = 3a³.

But I have not been able to find a single instance where this is so interpreted. The fact is that the rule requiring the operations of multiplication and division to be carried out from left to right in all cases, is not followed by anyone. For example, in case an indicated product follows the sign ÷ the whole product is always used as divisor, except in the theoretical statement of the case” (emphasis in original).

The following section of the text has the subheading “The Established Usage” reconfirms that the obelus indeed means the entire product after it is intended to be the divisor, and not just the first term. (Who says the obelus has no grouping power!) He also goes on to say “All multiplications are to be performed first and the divisions next.”

I’ve also found similar analysis in a 1935 textbook Second-Year Algebra by Hawkes, Luby, & Touton, p. 19. That’s how I understood it the late 70s and early 80s in high school and college and in the early 2000s when I taught JH & HS math and algebra. I seriously doubt nearly 100 years of history (at least) of this understanding has changed in the past ten years or so.

[2] Ambiguous PEMDAS (harvard.edu); see also https://math.berkeley.edu/~gbergman/misc/numbers/ord_ops.html

[3] Knill, Oliver. “Ambiguous PEMDAS.” Ambiguous PEMDAS (harvard.edu), retrieved May 29, 2023.