Chapter 4

Decisions, Decisions

Program Flow
I begin this chapter by introducing you to a new concept that you probably already understand intuitively. It is the concept of program flow. Every program traces a path as it executes, carrying out first one command, and then another, and another, until it reaches the end of the program; then the computer stops executing the program. Program flow is similar to the concept of a storyline for a movie: first the heroine does this, and then that happens, and then somebody else does something, and then the heroine does something heroic, and they all live happily ever after. The story moves from one event to another as it follows the storyline. So too does the computer move from one command to another as it follows the program flow. Perhaps we could even make a movie out of a single program’s execution:

Our story starts with FROGGER set equal to 4 ("Wow!"). Then, all of a sudden, he added 12 to FROGGER! ("Oh, no!") But, quick as a flash, he stored the answer into BIRDIE ("Whew, that was close!"). In the end, he printed both FROGGER and BIRDIE, and they were both correct. ("Yay!").

Granted, this would not make a very exciting movie, but it does show the similarity between program flow and storyline.

The important concept here is that the program flow starts at the first command of the program, moves to the second command, then the third, the fourth, and so on until it reaches the end of the program. The program flow follows a straight path from the beginning of the program to the end, just like a movie does.

At least, that’s how simple programs work. But in this chapter I am going to introduce you to something you will never see in the movies: program flow that branches. The best way to appreciate branching program flow is to imagine a movie that you could change. For example, imagine a movie in which the beautiful young heroine’s car mysteriously breaks down on a dark and stormy night in front of an old, abandoned house. Blood-red eyes peer from dark windows. "I’ll just wait in the house until morning", our innocent maiden chirps. "NO!" you cry out, "Don’t go into that creepy house!" But in she goes, the dummy. Well, what if the movie could change in response to your wishes? When you cry out, "Don’t go into that creepy house!" she pauses and says, " On second thought, I think I’ll just wait in the car."

If you think about it, you’ll probably agree that movies like this would not be as powerful or as interesting as regular movies; the heros and heroines would always do the mature, reasonable thing, and they would as dull and boring as the rest of us. But in this case, what makes a bad movie makes a good program. So let’s talk about branching.

Decisions
The essence of branching lies in decision-making. The act of making decisions carefully is one of the central components of Western civilization. We take it for granted, not realizing the degree to which we worship at the altar of decision-making. Perhaps a little story will drive home how crucial this religious devotion of ours is.

In 1757 India was a semi-autonomous collection of principalities under the partial domination of European powers. France and England contended for primacy on the Indian subcontinent. After a series of complex diplomatic maneuvers, the British commander, Robert Clive, led a tiny army of 3,000 men and 8 cannons against an Indian force of 50,000 men and 53 cannons. The odds looked bad, to say the least. The night before the battle, the Indian commander, Suraj-ud-Dowlah, celebrated the victory he felt certain to achieve on the morrow. Clive, by contrast, spent the night meticulously going over every aspect of the coming battle, inspecting troops, positioning his meager forces with painstaking care, planning for every possible contingency. When the battle was fought, the British forces emerged triumphant. The battle of Plassey delivered India into the British Empire. It was won, not by heroism, or superior firepower, or better discipline, or technological superiority, but by careful, thorough, meticulous planning. India was won by sheer force of decision-making power.

Precisely what is a decision? How do we make decisions? At heart, a decision is a choice between options. Three steps are required in the decision-making process:

1. Identify the options available
2. Identify criteria for choosing between options
3. Evaluate criteria and resolve the decision

An example might show this process in action. Suppose you are driving your car, about to enter an intersection. Suddenly an oncoming car makes a left turn directly in front of you. What do you do?

Let us begin by dispensing with all the nonsense options such as "Play a game of bridge" or "Whistle Dixie". Let us focus on the options that might avert an accident. Three simple possibilities come to mind: swerve to the right, swerve to the left, or hit the brakes and stop here. There might be more, but we don’t have time to debate; that car is coming fast!

Having identified our options, we must now identify our criteria for choosing between options. Our prime criterion is, of course, the avoidance of a collision. Given our uncertainties about the speed of the oncoming car, the condition of the road surface, and the intentions of the other driver, we cannot be certain as to the efficacy of any of the options, so we must think in terms of probabilities. Our decision criterion, then, is, "Which option has the greatest probability of avoiding a collision?" Again, we could get snazzy and throw in such considerations as the fact that an impact on the driver’s side of our car is more likely to injure us than an impact on the passenger side, but, again, let’s not dawdle on fine points when time is so short.

We gauge the relative positions and velocities of the two cars and assess the probabilities of avoiding collisions of each of the three options. From this assessment we conclude the following: if we attempt a straight-line brake to a stop, the probability of collision is very high; if we attempt to swerve to the right, the probability of collision is moderate; and if we attempt to swerve to the left, the probability of collision is low. Since our decision criterion is the lowest probability of collision, we choose the option "swerve to the left". That is an example of decision-making.

This example also demonstrates the fact that most decision-making is bedevilled by ugly complications and confusing uncertainties. These are the little nasties that demoralize our efforts to decide things carefully. We throw our hands up and declare, "It’s too complicated to figure out. I’ll just choose arbitrarily." It is sad to see people abandon their greatest human birthright, the ability to exercise free will by making decisions. There is another way to deal with complexity and uncertainty in our decision-making efforts. That way is to strip away the extenuating circumstances, to cut through the underbrush of complicating factors and get down to the heart of the matter. How can we reduce decision-making to its absolute essence?

The first step in this process is to reduce a complex decision to a series of binary choices. Instead of asking, should I take decision A, B, C, or D, we can use a process of elimination rather like that used in sports playoffs. To determine the best football team in the National League, we don’t throw all the football teams in the league into a single stadium and have them play one monstrous, twelve-sided game. Instead, we play a series of binary games, each game determing one victor and one loser, always pitting victors against victors, until there is but one ultimate victor. Choosing between two things is always simpler and easier than choosing between many things.

Sometimes we can simplify even further. When we are considering taking an action, sometimes it is possible to reduce a decision to "Do I take the action or don’t I?" This yes-no type of decision is the simplest possible way to approach decisions about actions.

In the case of a binary decision, the criterion for choice between the two options becomes ridiculously simple. It can be a simple true-false or yes-no type of criterion. In the case of football, the football game is a decsion-making process that determines the winner. If team A has the higher score, then team A is the winner. If team B has the higher score, then team B is the winner.

We have now arrived at a surprisingly simple formula for making decisions. It takes the form:

IF (condition) THEN (action)

"Condition" is something that is either true or false. There is no uncertainty or equivocation with "condition": it is one or the other. "Action" is the option that is selected. If this all sounds strange, just try the example: "condition" is "team A has the higher score". There’s nothing abstruse about that, is there? It’s either true or false; either team A has the higher score, or it doesn’t. And "action" is simply, "team A is the winner". Thus, the statement becomes:

IF (team A has the higher score) THEN (team A is the winner)

If you understand this simple concept of decision-making, then you are ready to use it in your programs, for this is exactly the way that decisions are made in BASIC. The statement in BASIC that makes decisions is called the IF-statement, and it looks like this:

IF condition THEN command

In this statement, the command-part is any regular BASIC command, such as FROGGER=5 or PRINT BIRDIE. The condition-part is a little trickier. It is a logical expression. A logical expression is not a command. It is a statement, a declaration that may or may not be true. Sometimes a logical expression can look just like a command, but that doesn’t make it the same. Here’s an example of what I mean:

10 FROGGY=5
20 IF FROGGY=6 THEN PRINT "YOU GOOFED!"

Line 10 is a command; it tells the computer, "Computer, I command you to put a 5 into FROGGY!" But in line 20, the reference to FROGGY is a logical expression. It asks the computer if FROGGY really is 6. The computer will, of course, evaluate the statement as false, because FROGGY is actually 5, and so it will not print "YOU GOOFED!"

Logical expressions can take many forms. They can evaluate equality with the equals sign (=), as in the above example. You can get much more complex than that example if you want:

30 IF (FROGGY+2)*5=(BIRDIE-7)/3 THEN PRINT "FROGGY’S the one"

For that matter, you can also determine if two numbers are unequal. The symbol to use for this is <>, a "less than" sign followed by a "greater than" sign. Together, they mean "is not equal to". So we could have the following code:

40 IF FROGGY=BIRDIE THEN PRINT"They are equal!"
50 IF FROGGY<>BIRDIE THEN PRINT"They are not equal!"

A particularly useful aspect of logical expressions is their ability to evaluate inequalities &emdash; whether one number is bigger or smaller than another. The symbols for these are ">" (greater than) and "<" (less than). I always get them straight by thinking that the big guy is on the big side of the sideways V. Their use is illustrated with these commands:

60 IF FROGGY>BIRDIE THEN PRINT "FROGGY is bigger than BIRDIE"
70 IF FROGGY


Just as with the equality symbol, you can get very messy with the logical expression:

80 IF ((FROGGY-7)*12)+3 > ((BIRDIE+4)/4)-8 THEN PRINT "What a mess!"

Now let’s lean back for a moment and consider broader issues. You may remember that I began this discussion by emphasizing the need to strip away petty details and get down to the heart of the matter. Now that we have reduced decision-making to its simplest form, we must now return to the question of making real-world decisions. All of this simplistic, yes-no decision-making may look great in a program, but what good does it do in the real world? It turns out that we can now use these simple decisions like building blocks to address a much more complex range of decisions. Mind you, we won’t be able to solve all the world’s problems, but you’ll be surprised at how much you can do with this simple IF-THEN statement. There are three ways to add richness to the decision-making that we can do with the IF-THEN statement: compound conditions, compound commands, and multiple options.

Compound Decisions


Sometimes you may want to consider several factors before taking an action. For example, suppose that your program is considering the age of the various people it is working with, but only wants to consider teenagers whose age is between 13 and 16? The program must consider two conditions: whether AGE is greater than 13, and whether age is less than 16. How do you put the two factors together?

In most BASICs you can solve this problem with Boolean operators. These are simple words, "AND" and "OR", but they are used much more precisely in BASIC than in English. You use these operators to couple two logical expressions, thereby creating a third, compound logical expression. The general rule is pretty much common sense:

AND: If you AND two logical expressions together, then the result is true only if both expressions are true (e.g., if expression A AND expression B are both true).

OR: If you OR two logical expressions together, then the result is true if either expression is true (e.g., if either expression A OR expression B is true.). If both expressions are false, then the result is false.

Suppose, for example, that AGE has a value of 14. Here are some examples of true and false statements:

StatementValue

AGE > 13 TRUE
AGE < 16 TRUE
AGE > 99 FALSE
(AGE > 13) AND (AGE < 16) TRUE
(AGE > 13) AND (AGE > 99) FALSE
(AGE < 16) OR (AGE > 99) TRUE
(AGE > 16) OR (AGE > 99) FALSE
((AGE > 13) AND (AGE < 16)) OR (AGE > 99) TRUE

This last example demonstrates the use of parentheses with logical expressions. If it generates confusion, just think of the parentheses the same way that you think about parentheses when you calculate numbers: figure out the innermost parentheses first and work outward. In fact, it’s quite practical to think of logical expressions in much the same way that you think of arithmetic, only instead of working with numbers, you are working with just plain old true-false answers. Let’s try it with the last expression in the list. First, look at each inequality and determine whether it is true or false; substitute that result into the expression. That gives:

((TRUE) AND (TRUE)) OR (FALSE)

Now discard unneeded parentheses that mark off single expressions:

(TRUE AND TRUE) OR FALSE

Now let’s collapse that TRUE AND TRUE phrase into a single result. You will recall that the rule for AND is that when both expressions are true, then the result is true. And indeed, both are true. So now we have:

(TRUE) OR FALSE

Discard the unneeded parentheses:

TRUE OR FALSE

And now remember the rule for OR: if either one or the other expression is true, then the result is true. Therefore, our expression works out to be:

TRUE

The value of compound expressions is that they make it possible to evaluate very complex situations. For example, consider this bit of code that one might see in a football program:

70 IF (DOWN = 4) AND ((YARDS > 2) OR ((THEIRSCORE-OURSCORE) < 20)
THEN PRINT"PUNT!"

Can you figure out what it means?

Compound Actions


What happens if you want to execute more than one command if the condition of an IF-THEN statement is satisfied? Suppose, for example, that you have a program that asks the user to input the age of a certain person, and you want to check to make sure that the number that the user types in is reasonable. If the inputted number is unreasonable, you want to print a message telling the user that and ask for him to input the value again, then thank the user for being patient. Your program might look like this:

50 INPUT AGE
60 IF (AGE < 1) OR (AGE > 99) THEN PRINT "That age is odd; please repeat."
70 INPUT AGE
80 PRINT "Thank you for your patience."
90 (this is the rest of the program)

This code would not do what you want. Even if the age were correct, it would ask for the age a second time, without explaining, and thank the user. In other words, lines 70 and 80 are executed regardless of the results of the IF-THEN statement. So there is the problem: how do you put multiple lines inside the "THEN" part of an IF-THEN statement?

The answer relies on a new command that is very simple in operation. The command is called GOTO and to use it, you simply type "GOTO n", where n is the line number of the line you want the computer to go to. The GOTO command breaks the normal program flow. As you remember, the program flow starts with the first command in the program and continues to the next largest line, then the next, the next, and so on until the computer reaches the end of the program. But the GOTO statement commands the computer to jump to whatever line number is specified. If you tell the computer to GOTO 60, it will immediately jump there and continue computing from line 60.

This is a very powerful command. It allows you to create whole groups of commands and execute them all just by telling the computer to GOTO the first command in the group. At the end of the group, you can tell the computer to GOTO the line from which it had earlier come. For example, a correct way to solve the problem of the bad age input would look like this:

50 INPUT AGE
60 IF (AGE < 1) OR (AGE > 99) THEN GOTO 1000
70 (this is the rest of the program)
1000 PRINT "That age is odd; please repeat."
1010 INPUT AGE
1020 PRINT "Thank you for your patience."
1030 GOTO 70

This code will execute properly; if the age is wrong then the program will jump to the corrective code in lines 1000-1030, then when it is done it will jump back to line 70 to resume the normal program. If the age is OK, it will do nothing in line 60 and go straight on to line 70.

There is one minor problem with it, a technical detail. What happens when the computer reaches the end of the program? It will eventually work its way through all the line numbers higher than 70 and come to lines 1000-1030. Then it will execute those lines, even though there was no error. Oops! The solution to this is a minor command that I never bothered to mention before. It is called "END" and it means just that. When the computer reaches the END statement, it stops executing the program. So, you should always put an END statement after your regular program but before all the little chunks of code like the example. In our example, the END statement might look like this:

200 END

As it happens, there is a neater way to solve the problem of the bad age input, but it requires a backwards approach. The code for it looks like this:

50 INPUT AGE
60 IF NOT ((AGE < 1) OR (AGE > 99)) THEN GOTO 70
62 PRINT "That age is odd; please repeat."
64 INPUT AGE
66 PRINT "Thank you for your patience."
70 (this is the rest of the program)

This code reverses the logic of the IF-THEN statement by using the NOT-operator. This operator just takes the opposite of a logical expression. If an expression is true, then NOT-expression is false. If the expression is false, then NOT-expression is true. It’s a simple idea, but it can be confusing to figure out. The same thing is true in English: no statement that isn’t written with no negatives is not knotty to figure out, no? Nevertheless, if you dig through it diligently, you can figure it out. Line 60 now says, in effect, "if the opposite of the old condition is true, then skip over lines 62-66 to line 70." The advantage of this approach over the earlier version is that lines 62-66 automatically feed into line 70 when they are done without requiring another GOTO statement. This approach cuts down on the amount of "spaghetti code" that you create. Spaghetti code is code that is full of GOTO statements, jumping all over the program. Programs like this are very confusing to read, and so they are difficult to work with. Since this latter approach is cleaner than the former approach, it is considered superior. However, please remember that both approaches work just as well. The difference is one of style, not function.

Multiple Options


Now we tackle the toughest problem: how do you put together all of these binary IF-THEN statements to handle complicated sets of options? For example, how would a BASIC program handle the traffic collision problem that I used at the beginning of this chapter? Let us assume that the computer is driving the car (Lord help us!) and is capable of calculating the probabilities of collision for each of the three options available. How could it choose among three options with its IF-THEN statement?

The answer requires the use of several IF-THEN statements in a sequence. Let’s say that it has stored the probability of avoiding a collision by swerving to the left into the variable LEFTSAFE. Similarly, it has stored the probability of avoiding a collision by swerving to the right into the variable RIGHTSAFE, and simililarly for STRAIGHTSAFE. It must decide which probability is highest and indicate the proper course of action. It does this by examining each in turn. Here’s how it’s done:

50 IF LEFTSAFE > RIGHTSAFE THEN GOTO 110
60 IF RIGHTSAFE > STRAIGHTSAFE THEN GOTO 90
70 PRINT "GO STRAIGHT"
80 GOTO 200
90 PRINT "SWERVE RIGHT"
100 GOTO 200
110 IF LEFTSAFE < STRAIGHTSAFE THEN GOTO 70
120 PRINT "SWERVE LEFT"
200 END

If you trace the program flow for each of the three possible cases, you will find that the program correctly deduces the correct course of action in each case. There is an ambiguity when two of the three values are equal; in this case, the program will prefer straight over right and right over left; that is an arbitrary aspect of the order in which the statements are executed.

Conclusions


This has been a long and involved chapter. What does it all mean? The important lesson here is that the computer really can make decisions. They are not the sort of soul-searching, agonizing decisions that we humans make; they are not expressions of free will. They are simple, mechanical decisions. Yet, we should not minimize the value of this kind of decision-making. We can differentiate between decisions involving incalculable factors and decisions that are merely complex. Moral and emotional decisions like, "Should we get married?" fall into the incalculable category, but many other decisions, such as "Which computer should I buy?" are at least theoretically calculable.

You might be surprised to know just how many decisions really can be submitted to calculation. The crucial element is our ability to express seemingly incalculable factors in quantitative form. Part of the problem is our own squeamishness with numbers, a reluctance on our part to reduce flesh-and-blood issues to numerical form. It seems dehumanizing to reduce issues to mere numbers.

But what is so "mere" about a number? Precisely what is wrong with expressing ideas in numerical form? Let’s consider a specific example. I once wrote a program (Balance of Power) that concerned geopolitical conflict. One of the variables I used in the program is called "INTEGRITY". As you might guess, this variable keeps track of the integrity of the player in his dealings with other nations. Now, integrity is one of our most cherished human virtues, something we revere as special, magic, beyond the soulless world of numbers. But is it not a quantity that a person can possess more or less of? Are not some people distinguished by great integrity, while others are possessed of little integrity? Is it not a small jump from "great or little" integrity to "100 or 20" integrity? Does not the numerical form allow us greater precision?

I will press the arguement even further. My program uses another quantity called SHOULD, which measures the degree to which the player should help another nation in time of need. The value of SHOULD is derived from treaty commitments made by the player. In other words, if I have made solemn treaty commitments to you, then SHOULD will have a high value, whereas if I have made only minor assurances, not guarantees, then SHOULD will have a low value. My program has a section dealing with the ramifications of a failure to honor one’s commitments. One command from that section looks something like this:

2240 INTEGRITY=INTEGRITY - SHOULD

Talk about blasphemy; here is a formula for integrity! Before you cross yourself and reach for the garlic, though, consider this: line 2240 does not present the formula for integrity; it presents a formula for integrity. The idea expressed in line 2240 is simple: if you fail to honor your commitments, then your integrity falls in proportion to the solemnity of the commitment. Is this not a reasonable concept? Does it not reflect the truth of the world?

Numbers and formulas are only a way to express ideas. They are another language. There is nothing intrinsically blasphemous about a language. Ideas can be blaspemous; had I written

2240 INTEGRITY=MURDER + LIES + THEFT

that would be blasphemy. The blasphemy does not arise from the quantification of the relationship, but rather from the relationship itself. Integrity does not arise from murder, lies and theft &emdash; no matter how you say it.

If you can learn how to express thoughts in quantitative form, you will have made a large step into a new world. Decisions become much clearer when they can be expressed in calculable form. But beware of taking your new-found skills too seriously; ultimately, all such decisions should be checked against simple common sense. Only a fool would take the equation in line 2240 as final truth.