“None of this will be important if there’s a zombie apocalypse. But how likely is that?”  —Jason Vladescu

I was once at a conference where I overheard a conversation at the next table during lunch. The two of them were talking about PSAs. I quickly realized that they weren’t talking about Public Service Announcements. They talked for about five minutes, though, before they realized that one was talking about his research into Pressure Swing Absorption while the other was talking about his research into Pressure Sensitive Adhesives. That conversation is the only thing I remember from that conference.


Semantics is the study of the meaning of words. Non-linguists tend to use the word dismissively, as in “you’re just arguing semantics.” While arguing over what words mean may seem trivial, agreeing on what words mean is essential. When someone using a word means one thing, and someone else hearing the word believes it means another thing, any agreement they achieve will be an amazing irony. Irony is not a good basis for process safety.

I don’t want to argue semantics. While some may want to argue about what a word is supposed to mean, I just want to make sure that when I’m in a process hazard analysis (PHA), the words that I’m hearing mean what I think they mean, and that the words that I’m using mean what I intend for them to mean.

Two of the most commonly used words in the field of process safety, especially in process risk, are “likely” and its cousin, “credible.” It always seemed to me that their meanings were obvious, until it became obvious to me that they meant different things to different people.

Likely, Unlikely, and Likelihood

The term “likely” is about probability. Probability, as a mathematical term, refers to the ratio of favorable cases to the whole number of cases possible. By definition, probability is greater than zero, and less than or equal to one. Or, if you prefer your probabilities expressed as percentages, greater than 0% and less than or equal to 100%.

Generally, “likely” means that the probability is greater than 50%, while “unlikely” means that the probability is less than 50%. Some want to separate the two terms with a grey zone in between, so that “likely” means a probability greater than, say, 75%, while “unlikely” means a probability less than, say, 25%. These are perfectly reasonable meanings, but not anything that is universally understood to be the definitions of the two words.

A word we often use for a probability of 100% is “certain”. A word we often use for a probability of 0% is “impossible”. This suggests that any probability greater than 0% is “possible” and that any probability less than 100% is “uncertain”, but I know that I don’t equate “possible” with “uncertain”.

While “likely” and “unlikely” refer to probabilities, during a risk assessment, “likelihood” almost always refers to a rate or frequency. Initiating causes are described in terms of failure rates. This is because in the absence of a time frame, probabilities are meaningless.  Don’t think so? Then consider the question, “What is the probability that you are going to die?” It’s 100%, which is not a particularly useful answer. A much more useful answer is to the question, “What is the probability that you are going to die in the next 12 months?” That’s what your life insurance company wants to know. Putting a probability in terms of a time frame transforms it into a rate or frequency, which is what I think of as a “likelihood.”

Possible and Impossible

For me, the words “possible” and “impossible” have nothing to do with probability. “Possible” means that the laws of nature do not prevent it from happening. “Impossible” means that the laws of nature do prevent it from happening—it is “not possible.” The second law of thermodynamics doesn’t say that perpetual motion machines are unlikely; it says that they are impossible.

Certain and Uncertain

The words “certain” and “uncertain” are not as related as they first appear. For me, “certain” means not only that the laws of nature do not prevent it from happening, but that the laws of nature require that it happen. When Little Orphan Annie sings, “The sun will come up tomorrow,” she is speaking to certainty. No wonder she’s willing to bet her bottom dollar.

“Uncertain,” on the other hand, addresses decision-making rather than scenarios, at least in risk assessments. For example, the toss of a fair coin has the certain probability of 50% for coming up heads, and the certain probability of 50% for coming up tails. With that scenario and those certain probabilities, I may still be uncertain about what to do in advance of the coin toss. It is the fact that neither “uncertain” nor “possible” are expressions of probability that make it so that they cannot be equated.

Credible and Incredible

Finally, there are the words “credible” and “incredible”. We use the word “credible” a lot, as in “That scenario is not credible.” We don’t say, “That scenario is incredible.” The word “incredible” is reserved for things that are unbelievable, or at least, difficult to believe.

Risk assessments reserve a very specific meaning to the word “credible”. In general usage, “credible” means believable. A credible witness is one a jury will believe, whether or not they are telling the truth. A witness that is not credible is one a jury will not believe, again whether or not they are telling the truth.

On the other hand, a credible scenario, at a minimum, is one that is not impossible. For some, that is enough, so “credible” and “possible” are equivalent. For many, though, the meaning of “credible” goes on to include that it happens at a sufficiently high frequency as to be worthy of concern. It doesn’t have to be likely, just likely enough, given a defined time frame. For these people, there is at least implied value of likelihood that separates the “credible” from the “not credible.” So, even if there is complete agreement on the nature of the meaning of “credible”, there is still some negotiation required to establish that dividing line.

Define Your Terms

It shouldn’t be necessary to argue semantics as part of a PHA. While it would be better if everyone in a PHA agreed on the meaning of these words—likely, unlikely, and likelihood, credible and not credible—it’s enough to make sure that everyone understands the meaning of the words being used. When you are facilitating a PHA, that’s your job.

Quantify these terms to the extent you can. You are working with technical people and without any guidance, each one will hang their own numbers on the words, so that the words don’t mean the same thing to anyone. Don’t be afraid to tackle it at the very beginning. Otherwise, misunderstandings can lead to bad decisions about process safety and process risk.

Finally, with apologies to Dr. Vladescu, an epidemic of brain-eating undead human corpses rising from the grave to attack the living is not unlikely. Given our understanding of human anatomy and death, a zombie apocalypse is impossible. It is not something you need to worry about in your next PHA.