Your Supervisor Asks What Changed Since the Last Draft

A supervisor's question about what changed since the last draft is not a request for reassurance. It is a request for a defensible record: which claims moved, which evidence arrived, which sections were cut, and what you now know that you did not know before. The fastest way to answer is to stop reconstructing the story from memory and start rebuilding it from artifacts you already have — the old draft, the comment thread, your search log, and your reference list.

This guide walks through that repair using a concrete case: a master's student returning in January to a thesis chapter that was last touched before the winter break. The same moves apply to a coursework paper, a first manuscript, or a new project start where the supervisor wants continuity rather than a fresh start.

Most students can describe what they did — read papers, rewrote a paragraph, ran a search. Fewer can describe what changed in the argument. The gap appears because drafting and evidence gathering usually happen in separate places: edits live in the document, searches live in a browser, and references live in a manager. When the supervisor asks for a diff, you have to reassemble three timelines into one.

The question also carries an implicit standard. A rigorous account of prior knowledge, strengths, weaknesses, knowledge gaps, transparent methods, and reproducibility is treated as core to research practice in NIH's scientific rigor guidance. Your supervisor is asking for a small version of that: not just "I edited section 3," but "here is the gap I closed and how I closed it."

Before drafting a reply, sort what you actually have. Most readers fall into one of three states, and each needs a different first move.

• You have a marked-up draft and a comment thread. Start from the comments, not the prose. Each unresolved comment is a change the supervisor already expects to see addressed.
• You have a new literature search but no written record of it. Start from the search itself and reconstruct what it returned, then compare against the references in the old draft.
• You have neither, only a vague memory of edits. Start by diffing the two draft files, then rebuild the evidence trail behind each substantive change.

The January thesis case sits in the second state: the student ran searches over the break, saved a handful of PDFs, and rewrote two subsections, but kept no log. That is recoverable, but only if the reconstruction happens before the reply is written.

Not every edit belongs in the answer. A useful filter is to report changes that alter one of four things: the claim being made, the evidence behind a claim, the scope of the argument, or the method described. Formatting, typos, and reordered sentences can be summarized in a single line.

For the thesis chapter, the student's edits sorted cleanly:

1. Claim narrowed. A broad statement about a mechanism was cut back to what the cited studies actually show.
2. Evidence added. Two new sources entered the literature review, one of which contradicted an earlier assumption.
3. Scope reduced. A subsection on a neighboring topic was removed to keep the chapter focused.
4. Method clarified. The description of how studies were selected was rewritten to state the eligibility criteria explicitly.

That fourth item matters more than it looks. Cochrane's guidance on searching treats planning searches from eligibility criteria, using multiple sources, combining free-text terms with subject headings, documenting the process, deduplicating records, and screening in stages as parts of a sound search. If your method section now says what your criteria were, that is a reportable change — and one your supervisor will likely welcome.

With the change categories settled, reconstruct the trail behind each one. Work from the search outward rather than from the prose inward, because the prose will tempt you to rationalize decisions you no longer remember making.

Open your search history and list the terms, databases, and dates. If the record is thin, rerun the key searches now so the log reflects a real, repeatable process. This is also where a scholarly search interface helps: running the queries again through WisPaper search gives you a dated, reproducible set of results you can point to, rather than a memory of what you think you found.

A common trap is treating a paper you had already cited as new. Compare the new results against the reference list in the old draft before claiming anything was added. Deduplication is part of a documented search process, and it protects you from overstating how much the evidence base actually grew.

If new references entered the draft over the break, verify them before they appear in your reply. BibTeX entries can carry errors or point to sources that do not exist. TrueCite screens BibTeX against real academic databases to flag hallucinated references, which is exactly the check to run before telling your supervisor that two sources were added.

A change log is the deliverable that answers the question directly. Keep it to one page and structure it so the supervisor can read the top line and stop, or drill into any item.

A workable format for each entry:

• What changed — one sentence, in terms of claim, evidence, scope, or method.
• Why — the gap, contradiction, or comment that prompted it.
• Where — section and paragraph.
• Support — the source or search that backs the new version.

For the thesis chapter, one entry reads: Narrowed the mechanism claim in 2.3 to match the two cited studies; removed the generalization that neither supports. Prompted by supervisor comment 4. Support: existing citations, no new source. Another reads: Added two sources to 2.4, one of which complicates the earlier framing. Support: January search, records verified.

Notice that the log records the reason, not just the edit. That is what turns a list of diffs into an account of your reasoning.

Once the log exists, the supervisor's follow-up is predictable: "Show me the new sources." If those papers are scattered across downloads and browser tabs, the conversation stalls. Consolidating them is a five-minute task that pays off immediately.

Save or upload the papers behind each change into My Library so the evidence for the new draft sits in one organized place. Then, when the supervisor asks what a specific new source actually argues, you can query the uploaded papers directly through Library QA instead of rereading under pressure.

This is also the moment to think about the next project. If the January cleanup is the start of a new chapter or manuscript, create it in Research Projects so the change log, sources, and drafts live together from the beginning rather than being reconstructed again in three months.

With the log and the organized sources in hand, the reply almost writes itself. A structure that works:

1. One-line summary of the most important change — usually the narrowed claim or the contradicting source.
2. The change log, pasted or attached.
3. Open questions you want the supervisor's view on.
4. What you plan to do next, so the reply ends with forward motion.

Lead with substance. "The main change is that I narrowed the central claim in 2.3 and added a source that complicates the framing in 2.4" tells the supervisor more than a paragraph about how much you read over the break.

The reply is ready when you can pass four checks:

• Every substantive change has a reason. If you cannot say why a claim moved, either find the reason or revert the change.
• Every new citation is verified. No reference enters the reply until it has been checked against a real source.
• The search is documented. Someone else could rerun your key searches from what you wrote.
• The log matches the draft. Open both side by side; a log that describes edits you later undid is worse than no log.

These checks echo a broader reporting discipline. PRISMA 2020 provides checklists and flow-diagram templates for transparent reporting of search, selection, and synthesis methods in systematic reviews. You are not writing a systematic review, but the principle transfers: if your method of finding and selecting evidence is transparent, your supervisor can evaluate the argument instead of interrogating the process.

The student's finished reply contains a one-page change log with four entries, a verified reference list, a documented January search, and two open questions about whether the contradicting source warrants a fuller treatment. The supervisor reads the top line, checks one citation, and responds to the open questions — a conversation about the argument rather than an audit of the process.

The transferable lesson is that the change log should be built while you work, not reconstructed afterward. A running log turns the supervisor's question into a thirty-second lookup. It also makes the next project start cleaner, because the evidence trail already exists.

Open the old draft and the current one side by side, and list every change that alters a claim, an evidence base, a scope decision, or a method description. For each, write one sentence of reason. That list is your change log — and it is the answer your supervisor is waiting for.