errors
Answers
-
What sort of errors please and where?
0 -
Search>1841 England Census
Last Name="Stretch"
Country="England"
County="Cheshire"
produces 148 results.
Somebody has interpreted the details recorded under "Residence" entirely incorrectly and shown them as places that are wrong, some of which are not even in the same country.
Some of the errors include:
"Cherry Lane" being shown as "Cherry Lane, Alleghany, North Carolina, United States "
"Aston Heath" being shown as "Estonia"
"Round House" being shown as "Round House, Navarro, Texas, United States "
"The Lodge Farm" being shown as "Farm, County Cork, Ireland "
"Steam Mill Street" being shown as " Street, Kent, England, United Kingdom "
There are others which I have not listed.
It seems that somebody apparently picked the result from a list and came up with a choice that was completely wrong.
0 -
The picking from a list is an automatic process that has gone horribly awry at times. That "somebody" was a programming error. The problem seems to be that if a partial place name is not in the database of standardized place names, the first thing that come up in that database was put for the place name. You can see how this works by testing these out in Family Tree. If you enter just Cherry Lane as a place name, look what happens:
Or Aston Heath:
This is a good cautionary tale. On a regular basis, someone will post a comment on this board along the lines of "Why doesn't FamilySearch just write a little program and fix everything at once?" Too many times they have tried that and had something like this happen.
This problem of automatic "standardization" going very wrong has been reported multiple times in multiple places involving multiple collections so the programmers have to be well aware of it by now.
You can sometimes get a sense of how extensive this problem is by looking at locations in a database to look for weird things. For example, the 1891 census for Norway has a number of errors due to this automatic process that backfired. The residence for every single entry in that census should be in Norway, but if I look at just people with last name starting with A and look at residences, there are are bunch not in Europe at all:
If I look at those in Europe, there are quite a few not in Norway:
So unfortunately, until they get this fixed, we need to keep in mind that searching by residence may miss the results we want and we need to ignore the Event Place name and only look at the Event Place (Original) name.
0 -
Same problem as above. So many errors re: place names mistranscribed and wrong county/country .
Family Search used to be a fantastic go-to for family history now is appears much of the information is incorrect - most frustrating.
0 -
On a percentage basis, very little of the information in the indexed records is incorrect. And in regards to the discussion here there is only one little mistake. Unfortunately it was a mistake that proves the saying,:
“To err is human but to really foul things up requires a computer.” —Paul Ehrlich, American scientist (1926- )
This simple mistake was to have the programming say, "If place name is a single level, standardize to the top entry in Places database list," instead of "if place name is a single level, do not standardize."
The correction is relatively simple: Delete all the standardized place names from the historical records, restoring them back to having just the Place Name (Original). Then fix the routine and try again.
A big part of the problem is that the Places database is so incomplete that I'm skeptical that any automatic routine will be able to standardize these places. To take my example from above, there is really nothing in the indexed record that tell you where Røken is, even if you look at the image it just says Røken Herrad. The way I would determine how to standardize this would be to go the Norwegian Archive's site and look for the same record:
Using the information on FamilySearch, it took me less than a minute to find it and see that Røyken is the current spelling of this municipality in Buskerud county. Checking the Places database, putting in Røken, the spelling on the census record and the correct spelling for 1891 I get:
Which shows where the error came from. But even if I put in the current spelling, the routine probably would have made a mistake because it seems to always take the first entry of the search results and here the correct standardization is the second entry:
I do hope they get this error corrected soon. One huge sweep through the database to reverse the actions of the attempted improvement and then figuring out how to redo things properly will be extremely helpful.
2