POL00333345
POL00333345
HSFPOLInquiryCorrespondence[;
Sent: Wed 10/01/2024 11:07:53 AM (UTC)
To: Smith, Laura (Post Office Horizon Inquiry).
Beaumont, Julie (Post Office Horizon Inquii
Naveed, Isra (Post Office Horizon Inquiry):
Nadeer i i i
_} Solicitor Post Office Horizon
Ce: Rowan Gregglt GRO HSF POL Inquiry
it OHorizonGold-Silver
Subject: Post Office Horizon IT Inquiry - Documents relating to Stephen Bradshaw [HSF-
LONDON_11.F1D3019466]
Attachment: 2024-01-05 - POL - Letter to the Inquiry - Documents relating to S Bradshaw.pdf
Dear Solicitor to the Inquiry,
Peters & Peters are assisting POL with the review of documents relating to Stephen Bradshaw. The content of this
email has been provided by Peters & Peters.
We refer to the note prepared by Peters & Peters which was provided to the Inquiry on Friday 5 January 2024
regarding the 942 documents produced to the Inquiry that same day. In that note, Peters & Peters explained that as
part of POL’s efforts to identify duplicative material to the Inquiry, KPMG had run a duplicate analysis over those 942
documents. Of the 942 documents, KPMG identified 420 as being either MD5# duplicates or 98-100% textual near
duplicates (TNDs) of documents previously produced to the Inquiry. However, Peters & Peters believed that a
substantial number of the remaining 522 “new” documents in the production also appeared to be TNDs of documents
previously produced to the Inquiry. These included ROTIs and investigation reports from Rule 9(14) Q42 cases which
have been produced to the Inquiry on multiple occasions.
KPMG has therefore investigated why those documents were not identified as part of its original duplicates analysis.
KPMG’s conclusion was that these documents were not flagged as TNDs because they fell below the requested textual
similarity threshold of 98%. KPMG explained that whilst the documents Peters & Peters had highlighted do appear to
be textually similar, small variations may have caused the documents to fall below the 98% similarity requirement.
KPMG therefore broadened its TND analysis from 98-99% to 90-99%. The results of that broader TND analysis
indicated that 578 documents in Friday’s production were either MD5# duplicates or 90-100% TNDs of documents
previously produced to the Inquiry.
KPMG’s most recent analysis indicates that Friday’s production contains 364 “new” documents. This remains contrary
to the feedback from Peters & Peters’ review team, who have indicated that a very large volume of the material from
Friday’s production was duplicative of documents previously produced to the Inquiry. Peters & Peters has therefore
begun a further manual review of the 364 “new” documents to determine whether on their face they appear similar
to material previously produced to the Inquiry. It is apparent from the preliminary findings of that manual review
that there is still TND of previously produced material within the 364 “new” documents. For example, Peters & Peters
has seen multiple POL legal memos regarding Hughie Thomas' court hearings, Hughie Thomas’ mediation application
and mediation report, and multiple copies of Hughie Thomas’ audit report, all of which have produced to the Inquiry
previously. POL has therefore asked KPMG to investigate these examples urgently to understand why these
documents have not been caught as part of its 90-99% TND analysis. In parallel, Peters & Peters is conducting a full
manual review of the 364 “new” documents from Friday’s production.
Following its further investigations, KPMG has been able to confirm that the additional examples were not identified
as TNDs as the duplicates previously produced to the Inquiry were processed by another firm and transferred to
KPMG. POL understands from KPMG that these duplicates contain ‘junk’ characters at the end of the extracted text
file, causing them to have significant textual differences to the documents in Friday’s Bradshaw-related production.
This means that, even though it appears that several of the items within the “new” 364 documents in Friday’s
production have been produced to the Inquiry as many as five times before, they have not been picked up in KPMG’s
duplicate analysis. This may explain why KPMG's analysis does not correspond with Peters & Peters’ view that the
POL00333345
POL00333345
"new" documents contain numerous duplicates. In addition, POL understands from KPMG that TND analysis is often
not effective when analysing Excel files due to the way text extracts formulas and content from the spreadsheet.
KPMG has confirmed that 44 of the 364 “new” documents from Friday’s production are Excel files. POL understands
that this might further explain why more potential duplicates have not been picked up as part of KPMG’s TND
analysis.
As POL’s further investigations will take some time to complete, POL yesterday arranged for a new overlay to Friday’s
production prepared by KPMG to be uploaded to Egress so that the Inquiry can see which documents have been
identified as duplicates as part of KPMG’s current 90-99% TND analysis. However, POL will endeavour to update the
Inquiry on the results of its ongoing efforts to identify further duplicates among the documents produced in Friday’s
Bradshaw-related production.
Yours faithfully
Herbert Smith Freehills LLP