Fact, Assumption, Hypothesis: A Beginner’s Mini-Glossary for Incidents

Rajesh Kumar

Rajesh Kumar is a leading expert in DevOps, SRE, DevSecOps, and MLOps, providing comprehensive services through his platform, www.rajeshkumar.xyz. With a proven track record in consulting, training, freelancing, and enterprise support, he empowers organizations to adopt modern operational practices and achieve scalable, secure, and efficient IT infrastructures. Rajesh is renowned for his ability to deliver tailored solutions and hands-on expertise across these critical domains.

When something breaks in production, people start using words loosely: “it’s the database”, “we know it’s the network”, “obviously a bad deploy”. Beginners often copy this style and then wonder why the team keeps going in circles. This primer gives you a small vocabulary for incident conversations, plus a small glossary to keep on a card, so that you can tell what is known from what is being guessed.

Why a vocabulary helps at all

Incidents are noisy. Messages arrive faster than anyone can read them, and many are partly remembered or half checked. A shared set of precise words lets a team compress a conversation: “that is a hypothesis” tells everybody the claim needs a test and must not yet be acted on. Without those words, each person decides privately how sure they are, and the group’s picture becomes a blend of different levels of certainty.

The mini-glossary

  • Symptom. Something you can observe going wrong, stated without a cause. “Checkout pages return an error for some users.”
  • Fact. A statement you have verified yourself, today, in a place others can check. “The error rate chart shows a rise starting at a given minute.”
  • Assumption. A belief you are relying on without having checked it, usually because it is normally true. “The cache layer is healthy.”
  • Hypothesis. A proposed explanation that predicts something testable. “If the new release is responsible, rolling it back in staging should remove the error.”
  • Evidence. An observation that supports or contradicts a hypothesis. A log line, a graph, a test result.
  • Root cause. The underlying reason, found by testing, not by consensus. It is the last item reached, never the first.

Copy those onto a card. Six words are enough to change the tone of a conversation.

Using the words on a real example

Imagine an online booking page that is loading slowly. A teammate says, “The database is overloaded.” As a beginner, you can respond with the glossary instead of an opinion. Is it a fact? Has anyone looked at the database’s load measurements in the last few minutes? If not, it is an assumption or a hypothesis. If someone says “I think the slowdown began after the evening release”, that is a hypothesis, because you can test it by comparing timing with the release time.

Notice how the glossary prompts action. Facts get recorded. Assumptions get a quick check. Hypotheses get a test with a predicted result. Nothing is dismissed; everything is routed.

A tidy way to write it down

Use a page with three short lists: what we know, what we are assuming, what we are testing. Each line gets a time and a name or role beside it. At the end of the incident, the page becomes the backbone of the review, and the contrast between early guesses and final facts is often the most educational part.

Groups that care about how people work together, not only about the tools, usually find this habit takes root faster. A guide on managing culture change when adopting DevOps describes how new shared habits spread through a team, and clear incident language is one of the easiest to start.

Common beginner mistakes

  1. Calling something a fact because a senior colleague said it. Seniority is not a source.
  2. Treating a recent change as the cause simply because it was recent.
  3. Running a fix before the hypothesis has produced a testable prediction.
  4. Forgetting to record the ideas that were ruled out.
  5. Changing two things at once, which destroys your ability to attribute the effect.

Everyone makes these. The point of the vocabulary is that each mistake becomes easier to spot and name in the moment.

How the glossary grows with you

Once these six terms feel natural, the rest of the field’s language is easier to absorb, because most advanced terms are refinements. Our wider longer glossary of starter-level DevOps vocabulary is a good next stop, and you will find that many entries turn out to be ways of describing what is confirmed, what is expected and what is measured.

A note on practice

Reading definitions is not the same as using them. Try narrating a small outage on a home project, or an old incident report, using only these words. If you want structure, the DevOpsSchool learning site offers one route: it describes self-paced courses with hands-on labs, plus certification tracks and study material, which makes a reasonable place to rehearse this vocabulary without a live outage.

What to say when you are unsure

Beginners worry that admitting uncertainty makes them look junior. In practice the reverse tends to happen. “I have not verified that yet, so I am treating it as an assumption” is the sentence of someone who understands the work. You can follow it at once with an offer: “I can check the dashboard now if that helps.” Pairing honesty with a next step keeps the conversation moving and earns trust quickly.

A quick self-test

Read these three statements and label each. One: “Our dashboard shows memory use at the top of its range on all three servers.” Two: “The new library must be leaking memory.” Three: “If we restart one server and memory falls but then rises again, the leak is likely in the application, not the host.” The first is a fact. The second is a guess that has not yet earned the word hypothesis. The third is a hypothesis with a prediction.

Remember this

Say “fact” only when you have seen it. Say “assumption” when you are leaning on something unchecked. Say “hypothesis” when you can say how you would test it. Do that consistently and your incident conversations will feel quieter, shorter and a good deal more useful, even when the problem itself is difficult.

Leave a Reply