Three documents, two versions, one month:
How the Bureau's language changed
In December 2025 I wrote about a linguistic study by Brooke Heller about four drafts of a Philip Morris speech, and about what the drafts showed: manipulation is not only what is added, it is what is erased.
http://www.elastictruth.com/2025/12/the-art-of-corporate-spin.html
http://www.elastictruth.com/2025/12/quantifying-deception-how-philip-morris.html
The word "addiction" was in the early drafts and not in the final one. References to a falling share price were removed. The text was shortened, and the sentences were made punchier, so the argument would be easier to accept without being checked.
This post does the same thing with three documents the Bureau of Meteorology published in September and October 2026. They are public. You can open them alongside this post and check every quotation.
The three documents are:
1. A letter to Andrew Wilkie MP, signed by Dr Stuart Minchin, CEO and Director of Meteorology, dated 29 September 2026.
2. A website statement, issued 22 October 2026.
3. A website statement, dated "Updated: 26 October 2026 | Issued: 22 October 2026."
Letter to Andrew Wilkie is here
Documents 2 and 3 are the same text. Only the date line differs. So there are really two versions, one month apart, and a reissue on the day a television segment aired.
Here is what changed between the first version and the second.
What was deleted
1. The name.
The letter of 29 September says:
"I would like to thank Mr Berger for the considerable effort involved in identifying and documenting the anomalies and patterns described in the report."
The website statement of 22 October says:
"The Bureau of Meteorology is aware of a report produced by a member of the public…"
Mr Berger is gone. There is no thank-you in the website statement. There is no name. There is "a member of the public," which is a category, not a person.
The same report. The same author. Two descriptions, one month apart.
2. The community.
The letter says:
"Contributions from members of the public and the broader scientific and citizen science community play an important role in supporting discussion and scrutiny of Australia's climate records."
The website statement does not contain the phrase "citizen science" at all. It does not contain the word "community." The audit is no longer a contribution from a community of people doing scrutiny. It is "a report produced by a member of the public."
3. The commitment to exclude.
The letter says:
"The affected records will be flagged and excluded until the original historical observations can be located and re-digitised."
The website statement replaces this with:
"We have corrected these values in this year's update, completed on 19 October 2026."
The first is a commitment to set the records aside until the originals can be found. The second is an announcement that the work is done. Those are different claims, and the second one is the one the files contradict: the duplicates were still present on 26 October and 28 October, and were removed on 29 October.
4. The word "negligible."
The letter says the effect is "negligible."
The website statement says the effect is "so small as to be statistically irrelevant."
"Negligible" is a judgement. "Statistically irrelevant" sounds like a measurement. The second is more authoritative, and it is doing the same work as the first. The word changed; the claim did not.
What was added
1. The completeness claim.
The website statement adds a sentence that is not in the letter:
"Further investigation has detected no further duplicate months in the entire bureau network other than those quoted in the report and then confirmed by us."
This is the most important addition in either document. It is a claim of completeness: not just "we found some," but "we looked, and there are no others."
It is also the claim that the files contradict. The audit identified duplicate blocks at Sydney that were not among the four months named. The website statement says no further duplicate months exist. The files show they do.
2. The uncertain cause.
The website statement adds:
"Historical climate records were digitised using methods including punch-card systems in the 1960s, which may have contributed to isolated transcription errors. The bureau is confirming if it was the cause in this case."
The letter does not mention punch-cards. The website statement introduces them, hedges them ("may have contributed"), and then concedes that the Bureau does not yet know whether they were the cause.
So the sentence reads: the errors may have been caused by punch-cards, and we are checking. A reader skimming the statement will remember "punch-cards" as the explanation. The document does not say that. It says "may have."
3. The pivot to staff.
The website statement ends with a paragraph that is not in the letter:
"The bureau is aware of commentary aired in the media regarding this topic, singling out bureau staff. While we respect that as an essential public institution, our work and information should be open to scrutiny, the bureau has zero tolerance for our people being individually targeted."
This is a change of subject. The document is about data integrity. The final paragraph is about staff welfare.
It is placed at the end, where it functions as a moral closing note, and it reframes the controversy as an issue of personal attacks on scientists rather than an issue of data.
That may be a legitimate concern. It is not a response to the audit.
What did not change
1. The reviews.
The letter says the reviews were in "2011, 2015 and 2018."
The website statement says the reviews were in "2011, 2015 and 2017."
One of the two is wrong. The reviews cited are real, but the dates differ between the two documents, and neither document says which review is which.
2. The structure of the argument.
Both documents use the same architecture, in the same order:
1. Acknowledge a small, bounded error.
2. Attribute it to historical technology or manual processes.
3. Quantify it as small using a denominator.
4. Dismiss the rest as "misunderstandings."
5. Cite independent review as validation.
6. Announce that the correction is done.
The wording changed between the two documents. The architecture did not. The letter and the statement are the same argument, written for two different audiences — one for a parliamentarian, one for the public — and the differences between them are the differences between those audiences.
What the changes do
Read the two documents as a sequence, and the movement is clear.
The letter is addressed to a named parliamentarian, on behalf of a named constituent. It thanks the constituent by name. It acknowledges a community of citizen scientists. It commits to setting the records aside until the originals can be found. It is a letter from an institution to a person, and it is written in the register of correspondence.
The website statement is addressed to no one in particular. It does not name the author. It does not mention the community. It announces that the correction is complete. It adds a claim of completeness that the files contradict. It adds a hedge about punch-cards. It ends with a paragraph about staff welfare.
The movement is from acknowledgement to closure. The letter says: we have found some errors, we will set them aside, and we thank the person who found them. The statement says: we have found some errors, we have fixed them, there are no others, and the person who found them is a member of the public.
That is the shift Heller found in the Philip Morris drafts. It is not a change in the facts. It is a change in the relationship between the institution and the person who raised the problem.
And then there is the date
The 26 October document is the 22 October document, reissued with a new date line. Nothing in the text changed.
It was reissued on the day the Media Watch segment aired.
Media Watch were on this this very website on that day too, the 22 October, despite them emailing on the 23 October saying the only information they can find on me was from a Speaker Bureau.
The effect of reissuing an unchanged statement on the day it is being discussed on television is to make it look like a fresh response. It is not a fresh response. It is the same response, with a new date.
That is worth noticing, because it is the same technique as the rest of the document: the appearance of movement, without the substance of it.
The question
The Bureau's statement says it has detected "no further duplicate months in the entire bureau network." The files show further duplicate months. The statement says the correction was "completed on 19 October 2026." The files show the duplicates were still present on 28 October and were removed between 28 October and 29 October.
Those two sentences, and the files that contradict them, are the whole of the dispute. The linguistic analysis explains how the sentences were written. It does not explain why the files say something different.
That is the question the Bureau has not answered, and it is the question the wording is designed to make the reader forget to ask.
No comments
Post a Comment