Doing a search by a person's age?
Can FamilySearch.org allow searches to be done based on a person's age? Right now, there are a lot of "false positives" being returned when one sometimes searches exclusively based on birth years and death years. For instance, if I want to find 108+ year-old Brazilians who died in 2001, I can put 2001 and 2001 in the death year range and 1880 and 1893 in the birth year range, as well as Brazil in the death place box, but this will give me an extremely massive number of false positives and almost no true results, which I suspect is inaccurate. But allowing searches to be done by age would allow me to see which, say, Brazilians died in 2001 and actually claimed to have been 108+ years old at the time of their deaths.
Comments
-
Most of those many results probably don't actually fit one of those parameters. The search inputs are not filters: they affect the rankings of matches, but if a record matches, say, "died between 2001 and 2001", then it will be somewhere on the list even if the birthdate is 1993 rather than 1893.
Unfortunately, FS's database is full of weird errors in dates, such as mid-20th-century deaths with a "death registration" field that has "0001" for the year, which the algorithm matches to "2001-2001". This means that not even using the actual filters results in a "clean" list (https://www.familysearch.org/search/record/results?count=20&q.deathLikeDate.from=2001&q.deathLikeDate.to=2001&q.deathLikePlace=Brazil&c.birthLikeDate1=on&f.birthLikeDate0=1800&f.birthLikeDate1=1800~1880&c.deathLikePlace1=on&f.deathLikePlace0=8&c.deathLikePlace2=on&f.deathLikePlace1=8%2CBrazil&c.deathLikeDate1=on&f.deathLikeDate0=2000&f.deathLikeDate1=2000~2000).
Given this existing mess, you can imagine how much worse it would get if FS were to attempt adding ages to every index entry. There are already problems where a misread or mistyped year or age results in an incorrect calculation that users can't fix and which doesn't get recalculated when its inputs are changed. I really don't think it would be worth the hassle to add such calculations more widely.
2 -
I just tried a similar exercise, but using ENGLAND instead of BRAZIL as the place of death. (Entering "England" as the birthplace is of no use, as death / burial records would rarely have a place of birth included in the record.)
As you can see, my search produced 6,123 results. As you say, these did not necessarily include results relating to the inputted data. The first results appear to at least be connected to a 1880-1893 birth, but there is a problem (that Julia references) with returning just 2001 deaths. Many death records have been incorrectly added to FamilySearch's database, so produce in inaccurate date. Take the example of Elizabeth Graham's record, in the second screenshot. The "1 Feb " Death or Burial Date is the field that has been used in the search algorithm, but the missing year has "standardized" as 0001. To further illustrate how complicated it is for a search routine to pick up on the requested range of dates, as you can see the Death Date field (obviously ignored in the exercise) shows a date of 29 Jan 1962!
In summary, at least for England records, it is of no use inputting the country name for the birthplace. And, for deaths, the search will generally "prioritise" the results - the "0001" ones appearing to be read as "2001", however! Whether FamilySearch programmers can reduce / eliminate these problems seems rather doubtful, given the mistakes / omissions in the records when they were added to the database.
1 -
I only checked out your illustration (via URL link) after my post, I'm afraid! I see you illustrated, in your example, the exact same point that you mentioned about "0001"in your text. Never mind, repetition has probably emphasised the nature of the problem in "achieving" what Daniel would ideally wish to see returned.
(The "sharp-eyed" will notice the Results figure differs in the two screenshots - I must have repeated the search and somehow "lost" 10 results in the process!)
1