The Description Trap

A recent LinkedIn post struck a nerve. I mentioned examining a thesis where the candidate had done excellent archival work: rare photographs, private family archives, oral histories with community elders who have since died. It’s irreplaceable material. And then presented it without really analysing it.

I received a number of messages that said the same thing: I think I'm doing this.

So it’s probably worth a closer look.

The discovery problem

Finding something nobody else has found is one of the best parts of research. A photograph in a private archive. A letter that changes the chronology. An oral account that contradicts the official record…. It’s amazing and one of the joys of research.

That excitement creates a problem.

When you uncover something remarkable, the instinct is to share it. To lay it out. To let the material speak for itself. So you write it up carefully and chronologically. X observed this in 1847. Y reported this in 1863. Z confirmed it in 1901.

The archive looks impressive. The effort involved is obvious. It’s important to set out the source material too. It helps us to understand and ‘get a handle on it’.

At that moment you've described your sources. You haven't strictly analysed them. Yes, you’ve put them into a sequence, or recognised a pattern, or linked disparate things together. It’s a good start - but don’t end there.

What analysis actually requires

A source is not a fact. It is a document produced by a person, in a context, for a purpose, with interests, biases, and blind spots. Your job is not to report what the source says. Your job is to interrogate it.

Why did this source say this? Every document was produced by someone with a reason to produce it. A 19th-century traveller's account of a port city tells you something about that city. But it also tells you about the traveller's nationality, commercial interests, assumptions about the people they were observing, and what their intended readership expected to hear. None of that context is in the document. You have to bring it.

What were the author's interests and blind spots? Colonial administrators wrote about merchant communities in ways that served administrative purposes. Missionaries wrote about local architecture through theological frameworks. Company records were produced to satisfy shareholders. None of these are neutral observations. The moment you treat them as neutral, you've stopped doing research and started doing transcription.

What does this source reveal that the author didn't intend? This is often where the most interesting work is. What a source omits, assumes, or glosses over can be as revealing as what it states. A report that never mentions women's spaces in a settlement tells you something about the report, not just the settlement. Silences are evidence.

How does this source sit in tension with the others? If two sources contradict each other, that contradiction is data. Don't resolve it by choosing one and ignoring the other. Ask why they differ. What does the disagreement tell you about the positions of the people producing each account?

The chronological arrangement problem

There is a version of the description trap that is easy to fall into and hard to notice once you're in it.

You have fifteen sources spanning three centuries. You arrange them by date. You write a paragraph on each one. The chapter builds a coherent picture of how understanding of your topic developed over time.

It feels like analysis. It has structure. It moves.

But it's letting the sources set the agenda. The sources determine what gets discussed and when. Your voice is editorial rather than analytical. You're curating a collection, not making an argument.

The test: could a well-informed research assistant have written this chapter? If yes, if the chapter would look roughly the same regardless of who wrote it, then it's descriptive.

Analysis produces something only you could have written. It requires a claim, a position, an argument that the evidence is used to support. The sources don't generate the argument. You do. The sources are what you use to defend it.

The problem with letting data speak for itself

This is not unique to archival research. It happens in qualitative interview studies, survey analysis, and practice-based research.

"The data shows that..." is one of the most common phrases in PhD theses. It is also one of the most misleading. Data doesn't show anything by itself. Data is inert until someone asks a question of it, frames it within a theoretical context, and draws a conclusion that goes beyond what any single data point supports.

If your findings chapter is substantially composed of long quotations followed by brief commentary, you have a description problem. "This participant felt that X" is not analysis. "These accounts collectively suggest that X operates differently in this context because Y" is closer.

Ask yourself, every time you finish a paragraph of findings: have I told the reader what this evidence means, or just what it says?

Why supervisors often don't catch this

Supervisors see drafts in progress. In early drafts, the priority is coverage: getting the material down, demonstrating engagement with sources, showing the empirical base is solid. At that stage, description is appropriate. You're building the foundation and conveying the data.

Some students never move off that foundation. The description appropriate at draft stage quickly becomes the submitted thesis. And a supervisor reading the same material for the fourth time is not always well-placed to notice the analytical layer never arrived.

Examiners come to the work fresh. They read your chapter without knowing how much effort went into sourcing the material. They ask: what is this chapter arguing? If the answer is "it's presenting the chronological development of X," that is not enough.

A practical check

When you finish a section, ask:

What is the claim this section is making? Not the topic. The claim. If you can't state it in one or two sentences, there may not be one.

Could I remove any of these sources without changing the argument? If yes, either the source shouldn't be there, or the argument isn't connected to the evidence tightly enough.

Am I telling the reader what the source says, or what it means? "X wrote that the settlement was prosperous" is description. "X's account of prosperity reflects the commercial interests of the trading company that funded his expedition, and should be read cautiously against local sources that suggest a more complex economic picture" is the beginning of analysis. Yes, it’s a bit forced here - but I’m trying to make the point quickly….

Have I acknowledged the limits of my sources? A source treated as straightforwardly reliable when it isn't will be picked up in the viva. Acknowledging limitations is good practice.

Description feels safe. You're reporting what's there. You can be challenged on your interpretation, but not on your transcription.

But a PhD is not a transcription. It's a contribution to knowledge that didn't exist before you did this work.

The archive is remarkable. The sources are valuable. Present them. Then tell us what they mean. That's the work.

Resources & Next Steps

If you're interested in temporary research/AI roles - I've been working with Mercor on positions for PhD students and researchers. Many of my readers have found these valuable for bridging funding gaps or gaining industry experience. If you're interested, applications take 15-20 minutes: https://work.mercor.com/?referralCode=6578187f-cfd6-41b3-84a4-0d34eeeac960&utm_source=referral&utm_medium=share&utm_campaign=platform_referral

Connect with me on LinkedIn: linkedin.com/in/iain-jackson — or simply reply to this email.

Please subscribe - I only write infrequently….

Thanks for reading. See you next time.

Iain