Re: Findmypast errors
- From: "Don Moody" <dpmoody@xxxxxxxxxxxxxx>
- Date: Fri, 30 May 2008 19:26:30 +0100
"Jeff" <jorg826@xxxxxxxxxxx> wrote in message
news:MEW%j.667$i74.393@xxxxxxxxxxx
eve@xxxxxxxxxxxxxx wrote:
This 1911 census is a chance to do it properly, not repeat the ill-
constructed shambles that was the 1901 census. If a commercial
firm is charging for access to an index, then it is up to them to
make as good a job of transcribing that index as is humanly
possible. It should not involve such known recipes for error as
employing non-Brtitish nationals, whether they be cheap labour from
the third world or American schoolchildren. You could forgive
freebmd if there were errors, since the service is provided free
and by volunteers,. Yet they go to a lot of trouble to get it
right. A commercial organisation ought to do just that, or not
take the money.
I totally agree.
In thee 1881 the LDS "normalised" counties etc which did improve
search capabilities (e.g. Oxon got changed to Oxfordshire and birth
counties were added when only a place name was provided.
I'd be interested in your views as to whether that is a desirable
pprocess. I'm not arguing either way, I'm just interested in your
views.
Any 'enrichment' of an index with searchable terms is 'desirable' but
there are two conditions which cannot always be met.
1. The terms have to have a constant, defined, precise, meaning.
2. The checking and amending has to be done to some very tiny level
of error.
In the case of county names, you'll get hit by condition 1. Names
change and borders change over the decades and centuries. To take a
large example, where is the county of Middlesex? There aint no such
animal any more, but several million people live within what were the
borders of Middlesex in my time as a lad, in my father's time as a
lad, and in my grandfather's time as a lad - all three of us having
experienced different 'counties'. What do my grandchildren do when
they come to search the index if they don't know (and they won't) that
Middlesex existed as an administrative entity?
In the case of condition 2, there is no way of ever proving that an
index is perfect. Roughly, costs of checking increase linearly with
each check, but the number of errors found decreases exponentially
with each check. Who makes the judgement of 'enough is enough' and
what does 'merchantable quality' mean in such circumstances?
As I've said on this group before, my outfit found by experiment that
17 checks were necessary for technical documentation to get it
apparently error-free. Since we were on terms of having to bear the
costs and consequences of any error, we were pretty damn careful. At
the end of it all we could only say that nobody had found an error -
yet.
Old hands in tecdoc know the story of the 707. In those days of paper
it was said to take two 707s to carry all the manuals to operate and
repair one of them. You would think the production was meticulous. In
true 'business' style it was put out to the lowest tender. Then a 707
lost its cargo hatch at 35,000 ft, explosively decompressed and hit
the ground. Nobody walked away. Hundreds died. The insurance claim was
$70 million that long ago. It would be a hell of a lot more now. The
cause was traced to the manual. The engineer writing it had imagined
the lock from where the works could be seen - inside the plane. From
there the locking mechanism involved a clockwise turn of a ring. But
the instructions were for the baggage handler outside the plane. So
when he turned the ring clockwise he was not locking the hatch but
unlocking it. The instruction for him should have read
'anticlockwise', but none of the few checks picked it up, and more
checks weren't affordable if a profit was to be made out of
lowest-price tender. Was the manual of merchantable quality if that
was the only mistake?
It isn't just the incidence of errors but the consequences of them
which need to be fed in to answering that question in respect of any
large assembly of information.
Don
.
- Follow-Ups:
- Re: Findmypast errors
- From: Keith Henry
- Re: Findmypast errors
- From: cecilia
- Re: Findmypast errors
- References:
- Re: Findmypast errors
- From: eve
- Re: Findmypast errors
- From: Jeff
- Re: Findmypast errors
- Prev by Date: Re: The enigmatic family of William Stewart Reed
- Next by Date: Re: Just getting started on the FAQ site...
- Previous by thread: Re: Findmypast errors
- Next by thread: Re: Findmypast errors
- Index(es):
Relevant Pages
|