Read on Twitter
1/22 Long thread: Imagine that BCCDC COVID-19 data is guarded like money in a bank (Bank of BCCDC). This bank has only 6 accounts: Fraser & Vancouver Coastal Health are largest, followed by Interior, Northern, Vancouver Island Health (smallest), & Unknown (last one a mystery).
2/22 You own the VCH account. For months, everything is normal. Occasional minor errors, quickly corrected. But there are rumors of mysterious transactions between accounts. You check paper statements & find everything intact. Except the online statements now differ from yours.
3/22 "Unknown" is accumulating a huge balance, transferred from other accounts. Some monies disappear only to be partially replaced. You talk to bank manager Reba Kristofferson. She assures you it's a computer virus. She will fix it. Her secret talent is talking to viruses.
4/22 This is how the “Bank of BCCDC” launders your data.
I’ve spent weeks scanning BCCDC .csv data for irregularities. No smoking gun. But I didn’t have archives from early summer, until after speaking to @JPSoucy. More files to analyze back to Aug 24. At last, daily downloads!
I’ve spent weeks scanning BCCDC .csv data for irregularities. No smoking gun. But I didn’t have archives from early summer, until after speaking to @JPSoucy. More files to analyze back to Aug 24. At last, daily downloads!
5/22 I put data on a spreadsheet and compared for the same line item as to Health Authority, Date & quantity of New Tests/Total Tests. What I found:
• “New Tests” & “Total Tests” numbers were retroactively changed. Not just days back, but months back after first uploaded file.
• “New Tests” & “Total Tests” numbers were retroactively changed. Not just days back, but months back after first uploaded file.
6/22
• On .csv data downloads in Nov/Dec, minor changes were observed in records from late Jan. Large changes were seen in records from Mar. This guy below says it all:
• On .csv data downloads in Nov/Dec, minor changes were observed in records from late Jan. Large changes were seen in records from Mar. This guy below says it all:
7/22
• Changes characterized by repeating number sequences were found from Mar through Dec for the same Health region each time and over consecutive days! Legitimate changes are inherently random and the numbers aren't identical. This ain't no Pee-Wee mistake!
• Changes characterized by repeating number sequences were found from Mar through Dec for the same Health region each time and over consecutive days! Legitimate changes are inherently random and the numbers aren't identical. This ain't no Pee-Wee mistake!
8/22
• “New Tests” & “Total Tests” changes are legit for minor errors, lost paperwork, etc. But some were huge. Numbers get added to Health regions, removed on a later download.
• Large numbers go from “Unknown” to a Health region, then often back to “Unknown”. Weird
.
• “New Tests” & “Total Tests” changes are legit for minor errors, lost paperwork, etc. But some were huge. Numbers get added to Health regions, removed on a later download.
• Large numbers go from “Unknown” to a Health region, then often back to “Unknown”. Weird
.
9/22
•The “Total Tests” for Unknown keeps growing.
Here's an example of a repeating number sequence: 7, 22, 5, 1, 9. “New Tests Δ” is the difference betw. the “New Tests” column on Sep 23 & the same column on Sep 15. “Total Tests Δ” is the difference betw. Sep 23 and Sep 15.
•The “Total Tests” for Unknown keeps growing.
Here's an example of a repeating number sequence: 7, 22, 5, 1, 9. “New Tests Δ” is the difference betw. the “New Tests” column on Sep 23 & the same column on Sep 15. “Total Tests Δ” is the difference betw. Sep 23 and Sep 15.
10/22 Compare lines 1564-1571 from the side-by-side downloads of Nov 19 & Nov 23.: There is a repeating pattern during Sep ( 8, 1, 4, 2, 8, 8) & Nov (-57, 1, 4, 2, -50, -15, 1).
11/22 Data moves get larger. Sum of the red numbers cancels sum of green numbers on any given day. This is how data gets reconciled. Bank analogy: moving money from one account to another, total cash in bank stays the same. For BCCDC data & banks, if no valid reason it's illegal.
12/22 “Corrections” continue up to Dec. Size of “New Tests” change gets larger – now up in the 1000’s. Is it reasonable to have that many “corrections” for one day? Vancouver Coastal seems to be getting most of its numbers transferred from “Unknown”. It doesn’t make sense!
13/22 Here is another repetitive pattern on Dec 2. Only somebody got clever & changed some numbers so they weren’t exactly the same. Why do VCH and FH get singled out for big changes? Hint:They have the largest school population and the largest number of MLA ridings
14/22 “Total Tests” number changes getting bigger at 14,000+. They are semi-repetitive with minor number tweaks. Prolly the work of this tin foil hacker that @BillTieleman knows.
https://twitter.com/RonaldNHughes/status/1340144054961467392
https://twitter.com/RonaldNHughes/status/1340144054961467392
15/22 Dec 13 data. More backdating to Mar. Same number (117) gets added to “Unknown” from VCH for Mar 13,14, 15, 16, 17. That's remarkable coincidence! Especially when “Known” data of VCH over months & months suddenly became “Unknown” on 5 consecutive days on Dec 13 data dump.
16/22 More repetitive numbers, this time bumping up Northern Health results in the “Total Tests” column. The “New Tests” number for a single day in Northern ranges between 349- 438. Yet the daily correction to Total tests is bigger at 2293. Whoa!
17/22 Dec 20 data dump, with more number repetition (7332). Total Tests changes as follows: 5082 added to VCH, 7332 added to “Unknown” grouping, 12615 subtracted from Fraser Health. These are *not* your typical minor corrections.
18/22 Dec 21, last day that I noted repetitive number sequences. “Total Tests” daily nos. increased by 397 for Fraser, 407 for VCH. No reconciling numbers by moving tests from “Unknown” to others. Tests just materialized from thin air – same number for 3 consecutive days!
19/22 Hey @RonaldNHughes: Bonnie Henry can really exceed 3000 daily tests. It's done with the magic of BCCDC spreadsheets! https://twitter.com/RonaldNHughes/status/1287135445826977792
20/22 So there we have it. Numbers repetitively inserted into downloadable databases to change the "New Tests" & "Total Tests" figures. The changes likely skew the daily positivity, but we'll never know because the positivity numbers aren't back-changed significantly.
21/22 We do know the numbers started changing in Aug , the end of Sep and near the end of Nov (will discuss in a future thread). This is borne out by huge number changes near those dates, all with unique pattern that appears to be repetitive and *likely human-generated*.
22/22 Will @adriandix investigate this apparently altered data as BCCDC falls under his Ministry? We know he isn't directly responsible, nor would he condone such illegality. Cause he's no longer an immature 35 y.o. backdating memos for Glen Clark. https://twitter.com/RonaldNHughes/status/1326659376787058688
For those asking, here is a Dropbox link to my spreadsheet. Data files that I used originate from archives stored on http://opencovid.ca https://www.dropbox.com/scl/fi/u0sr77tw6sunlsg8te7ch/Dashboard_Lab_Information_comparison-Aug-24-to-Dec-29_Dropbox.xlsx?dl=0&rlkey=0k7mvm3h0dcuj6e9d3jmfk6mz
