The danger of estimated dates

+12 votes
I just added a 12th century line of counts, none of whom has known birth dates. I did leave the birth date field blank and thus created quite a number of duplicates to profiles entered with estimated birth dates. Everything else identical, first names, family names, death dates and on ... Obviously, the program does not match even otherwise identical profiles if one is missing a birth date. But is entering some arbitrary estimate really a solution? These profiles span several generations and as noted previously in G2G discussions the error compounding over several generations can be quite dramatic making estimates not only useless but harmful. Perhaps we should rethink the emphasis on using estimates?
in Policy and Style by Helmut Jungschaffer G2G6 Pilot (549k points)
I experienced some problems when well-meaning individuals added estimated dates to undated profiles of Holocaust victims I was working on . The standard they used was dad born 25 years before kids, but in the case of Eastern Europeans, this was not good for the most part, as men in the 19th century there were often subject to compulsory 10 year military service and were forbidden to marry during that time. Thus, many Eastern European Jews didn't seek wives until they were about 30 and frequently I found, many lost their first born children in infancy or early childhood. So, dad born 25 years before the kid profiled, was a no-go most times and may have hindered other researhers on those lines looking for dad 10 years too early
Perhaps a closer inspection of the surname would have revealed those other profiles before you created the duplicates.  I wouldn't expect to search every John Smith born from 1880 to 1900 but I wouldn't think it unreasonable to conduct a basic search of a surname occuring in the 13th century.

The exact opposite argument could be made that profiles with no dates will be rejected as matches by WikiTreers who don't see any similarities between the profiles.  If a member creates a new profile for Joe Smith and they know that he was born in 1820 and your Joe Smith has no date at all, they are not likely going to select your Joe Smith as a match because he could have been born in the 13th century or in the 21st century.
We are not talking about profiles with no dates. If Joe Smith died about 1195 he should really not be confused with somebody from the 21st century.
From what I'm reading from in your post explanation it seems that you're saying that changing the search parameters would fix the problem?  If so then the argument is for a better search algorythm, one that also looks at death dates.
It would make sense, wouldn't it, since WT requires only one date, either birth or death.

I have exactly same issue - and i can confirm - estimating artificial information is very harmful and provides huge error from my copy of data in Myheritage and Geni. Instead would be much logical to keep dates = null, and have "Not living indicator". But apparently form has a bug and doesnt allow that. I raised discussion here:

1 Answer

+9 votes
I think it is important that a profile have a date in at least one date field (even if it's an estimate that could be off by 30-50 years) in order to try to show the general time frame for the profile. I feel less strongly about whether you need an estimated date of birth if you have a date of marriage and/or date of death, but I think any reasoned (and explained) estimate is better than nothing.

How could including an estimate be more harmful than leaving it blank?
by Chase Ashley G2G6 Pilot (245k points)
These profiles all had entries in the date of death field. And talking about the 12th century all dates we know are from surviving documents about known historic facts and people mentioned as participants or witnesses. The only thing one can often say for sure is that somebody was alive at a certain point in time and dead at another. When they first appear in the records they could have been 15 or 65, who knows? WikiTree is not designed to deal well with these situations, unlike specialized databases for this time period which give us the opportunity to use "before" and "after" without equating such dates just with the year mentioned, or allow time frames "between ... and ...", or the simple "mentioned". For most of the dates in question there is no reasonable explanation for an estimate except "I think so". And the harm lies in the potential for duplicates.
Currently, WT doesn't allow for the marriage date to replace the birth/death dates, which can be frustrating when it's the only provable date you have.  I'd like to see this changed.

Related questions

+4 votes
5 answers
+8 votes
2 answers
341 views asked Jun 29, 2018 in The Tree House by Dale Byers G2G Astronaut (1.5m points)
+6 votes
2 answers
+29 votes
7 answers
288 views asked Apr 25, 2018 in Policy and Style by anonymous G2G6 Pilot (129k points)
+7 votes
1 answer
+6 votes
1 answer
+4 votes
4 answers
184 views asked Feb 9, 2018 in Policy and Style by Jim Naureckas G2G Crew (800 points)
+9 votes
2 answers
+3 votes
5 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright