ElasticTruth

Update, 29 October: Media Watch says the record was fixed. 

The documents say otherwise.

Media Watch has written to Dr Jennifer Marohasy stating:

"We have followed up and have received word from BoM that it did in fact fix the record, and our program was correct when we went to air. It tells us that all ACORN-SAT locations, including Sydney Observatory Hill… had been updated as at the time we went to air."

The letter also says: "I suspect you are referring to the raw data, which was also in the process of being fixed, and I believe, should now be done."

Read those two sentences together. The first says the record was fixed at the time of broadcast. The second says the raw data was "in the process of being fixed" and "should now be done." 

Those are different claims about different datasets, and the second one is an admission that the work was not complete on 26 October.

That admission is the point. It is in Media Watch's own letter.


What the BOM letter says
The letter to Andrew Wilkie MP, signed by Dr Stuart Minchin, says:
"We have confirmed a small number of duplicated historical daily temperature records, including four post-1910 months relevant to ACORN-SAT. These account for fewer than 200 daily values from a dataset containing more than seven million values."

The statement of 22 October says the same thing:
"These were 4 post-1910 months relevant to the Australian Climate Observations Reference Network – Surface Air Temperature (ACORN-SAT)."
"We have corrected these values in this year's update, completed on 19 October 2026."

Both documents name four months. Both say the correction was completed on 19 October.


What was actually "corrected", read deleted
On 26 October, the day the segment aired, I checked the public CDO files. The following were still present:

- August 1867 / August 1907, Sydney — the front-page example. Still an exact copy.

- June 1943 / June 1944, Sydney — thirty identical days, both flagged Quality = Y. Still present.

- December 1935 / December 1936, Sydney — the seam, with fourteen identical days, a two-day insert, and a malformed record at the join. Still present.

- January / February 1910, Sydney — the seven-day block copied across the month boundary. Still present.
On 28 October, I wrote to three parliamentarians stating that the records had not been corrected.

On 29 October, the records were corrected. The June 1943/1944 replication, the December 1935/1936 seam, and the January/February 1910 block are now de-duplicated in the public file. 

So are the Charleville maximum and minimum records.
That is more than four months. 

The June 1943/1944, December 1935/1936, and January/February 1910 blocks were not among the four months named in the BOM letter or the 22 October statement. They were corrected on 29 October, after the segment, after the blog post, and after the emails to parliamentarians.


The timeline
- 19 October — the date the BOM says the correction was completed.

- 22 October — the BOM statement issued, saying the correction was completed on 19 October, and naming four months.

- 26 October — the segment airs. Media Watch says the duplication had been "swiftly fixed." The duplicates are still in the public file. I publish the files.

- 26 October — Dr Marohasy writes to Media Watch identifying the 1867/1907 duplicate as still present.

- 28 October — I write to three parliamentarians stating the records have not been corrected.

- 29 October — the records are corrected, including at least three blocks that were not among the four months named in the statement.

The two claims that cannot both be true
Claim 1: "We have corrected these values in this year's update, completed on 19 October 2026." (BOM statement, 22 October)

Claim 2: The June 1943/1944, December 1935/1936, and January/February 1910 blocks were present in the public file on 26 October and were corrected on 29 October. (The files, dated)
The statement says the correction was completed on 19 October. The files show three of the blocks named in the audit were still present on 26 October and were corrected on 29 October. The statement named four months; the correction covers at least seven.
Media Watch's letter says the program "was correct when we went to air." The letter also says the raw data was "in the process of being fixed" and "should now be done." Those two sentences are in the same email. 

The second is the one that matters.


What I am asking
I am not alleging fraud, and I am not asking anyone to accept my interpretation of the record. I am asking the Bureau to answer three specific questions:

1. On what date was each correction applied to the public CDO file?

2. Why did the 22 October statement name four months, when the correction covers at least seven?

3. Why did the statement say the correction was completed on 19 October, when the public file on 26 October still contained the June 1943/1944, December 1935/1936, and January/February 1910 blocks?
The dates are checkable. 

The files are public. 

Anyone can verify the sequence.

The Bureau has said the impact is "statistically irrelevant." That may be true of the four months it named. It is not a description of the correction that was actually applied.


On the "process of being fixed"
Media Watch's letter concedes that the raw data "was also in the process of being fixed" on 26 October, and that it "should now be done."

That is the admission. The segment said the duplication had been "swiftly fixed." 

The letter, written three days later, says the fixing was still in progress on the day of broadcast.

The correction was completed on 29 October. 

The segment aired on 26 October. 

The statement was issued on 22 October and said the work was completed on 19 October.

Three dates, three different claims. The files are the record, and the record is public.

update from: http://www.elastictruth.com/2026/10/bom-claim-of-fixing-problems-from-audit.html