Why does FamilySearch Places always suggest the newest place name

Rolf Ketil · April 26

In 99% of cases in genealogy research, we work with data that is older than 75 years because the archives are closed to those of us who are not immediate family. We do not have access to such data, for example in Vestlandet in Norway and the Innlandet in Norway.

Shouldn't it be the other way around where the oldest place names come first when newer source data in the archives is blocked from view during our lifetime?

Jack Hern · April 28

When you review and update your locations to standard place names; I like to scroll through to find the appropriate time period for the relative. With many choices, it is important to take a look. Different countries will be further along on the addition of historical time period place names; until then use the standard, and place your historical full name in the reasons box below for anyone to see if they so choose.

Another reason to take the time - the more complete, and standard, you can enter your vitals - the more and better the search algorithms will work for you. I.E. complete names, full dates, full locations; standard formatting. Nicknames and name variations placed in their proper space in 'Alternate Names', not in the primary Name field.

Rolf Ketil · April 30

This is not an answer to my question. You are only describing a general method for finding the correct standard place name.
My question concerns the actual indexing of the database. Why does the compiler suggest a standard place name for the 2020s, when genealogy is about old historical data? It takes time to scroll and if you are not careful, the place name will not correspond to the period in which the person lived.

Ascending order is the opposite, something a computer engineer understands, not a genealogist.

Paul W · April 30

In general, I can find no sense in the way place name variants are ordered in the drop-down.

As you say, the order generally appears to start with the most recent standard name for a place, but the issue is not as straightforward as that. I tested with three examples of placenames.

Firstly, Whitley Bay. There appears to be only one location of that name in the database, so your point is clearly illustrated here: three choices, starting with the format from 1974, then as appropriate for 1801-1974 and lastly for the period up to 1801. I agree with your argument here that the programmers should have set a procedure so these choices appear in reverse order.

However, in the case of "Barking" and "Corringham" the list appears to be ordered quite randomly. Okay, it's moving away from your main point, but why (and this is quite usual in the instances I encounter) does the list start off with a place in Essex, followed by two places in Lincolnshire, then continues with an Essex option, two choices for Lincolnshire and finally a further option for the place in Essex?

"Barking" is even worse, as the order of the list jumps between places throughout the globe that begin with "Barking…." - including "Barkingside" (one word) before continuing with more actual "Barking" placenames!

I believe this clearly illustrates there has been no serious programming applied to this feature, whether in ordering time periods or totally different locations. You will also find there is a total inconsistency in how a parish name is formatted - so, for example, if there are two or more parishes in a town / city, sometimes the parish name (say "St Margaret") will appear before the name of the town, but in other cases the town name will appear first, followed by the parish (church) name.

I believe the far wider issue of how these choices are currently presented needs to be examined, with many options being removed altogether. For example, having an identical placename as both a "parish" and a "town" makes no real sense, especially when it means (with very common names, e.g. Oxford) the option we are seeking does not even get to be included on the list without us specifying "which Oxford" we are wishing to use (however far we "scroll down").

Example of what is being found by the OP. I agree, no way should the most recent / least likely to be used format appear at the top:

Adrian Bruce1 · April 30

To be quite honest, I don't think that oldest first will always, on average, be better. Consider places in the USA that have not changed their higher level jurisdictions since 1776 - or not long after. Given the large number of immigrants late in the nineteenth century, presenting placenames ending in British Colonial America (ie pre 1776) first is quite likely to be frustrating to someone whose relatives entered the USA about 1900. Ditto if somewhere now in Nevada is presented with Utah versions for someone arriving in the state in the 1900s.

Conversely, if I look at my Virginian relatives of relatives, I would prefer the older Virginia names first in the list instead of the later West Virginia names.

So it's a bit all over the place. Perhaps the optimum would be to be able to sort the list and have that sort order be sticky.

As for the weird ordering of Barking, etc, I am certain that the order is telling us something about the algorithm - I'm just not certain what that something is, because I've no idea what the algorithm is. I'm not even certain that it's a conventional algorithm as distinct from an AI style (?) pattern recognition.

MandyShaw1 · April 30

Yes, it would be great if FS could document how this mechanism works for us. (Perhaps they might be appearing in reverse order of overall record count?)

MandyShaw1 · April 30

I've done some poking around. This page:

https://www.familysearch.org/en/developers/docs/api/places/Places_Search_resource

is informative, it says the results are provided in "best guess first" order, whatever that may mean.

I searched for 'Bath', which has 115 results according to https://www.familysearch.org/en/research/places/?pagesize=100&text=bath

The order of the entries (which is the same everywhere, including the Places Search API, https://api.familysearch.org/platform/places/search?q=name:Bath&count=100

appears to be controlled by:

First, a Score value (descending, not shown on the UI) - presumably this indicates the 'best guess' as mentioned above, but unfortunately there are only 6 different scores across the first 100 entries.

Second, the number of Jurisdictions in the hierarchy for the place (ascending).

Third, the place id number (ascending, not shown on the UI).

So this is what we get (first 100 only): [still preparing the table]

MandyShaw1 · May 1

I've now realised that a name search and a partial name search give slightly different ordering - will have another go at this later (out today).

Rolf Ketil · May 1

The reason I originally wrote this post is because we who do genealogy research are searching for our relatives who have passed away. The most important source that we want to document comes from original documents that are usually archived sorted out by parish or municipality. I personally am best versed in Scandinavian sources.

Now it happens that some political committee up through the ages has changed the parish from one municipality to another and even worse from one municipality to another county.

We are searching for historical data and for those who do not know local history will always struggle to find the source for the person's parish and our time here on earth is limited in terms of time.

When it comes to Scandinavian sources, more and more sources are being scanned, transcribed and published via the national libraries and sorted under their original parish. In Scandinavia we have several parishes and municipalities with the same name.

So, where should someone who is not familiar with the locality look? Standard place names must therefore become more correct in relation to their periods throughout history. This is historical data from ancient times, not the present, and should be sorted accordingly.

MandyShaw1 · May 2

I've got to the bottom of the sequencing, I think (though not the rationale behind the score, that remains a mystery).

Score, descending.
End date, descending ('current' places first).
Number of jurisdictions, ascending.
Unique identifier of place, descending.

Would a mod be able to get confirmation from the engineers that this is the sequencing used, and also a statement of how the scoring algorithm works, please?

Here's the Corringham one (partial name search, which the drop-down lists seem to use).

rank	score	displaytype	to-date	identifier	displayfullName	numJurisdictions
1	77	Town		2970449	Corringham, Essex, England, United Kingdom	3
2	77	Parish		2966284	Corringham, Lincolnshire, England, United Kingdom	3
3	77	Hundred	1880	3039970	Corringham, Lincolnshire, England, United Kingdom	3
4	69	Town	1801	10605332	Corringham, Essex, England	2
5	69	Hundred	1801	10602659	Corringham, Lincolnshire, England	2
6	69	Parish	1801	10602655	Corringham, Lincolnshire, England	2
7	61	Cemetery		11746269	Corringham Cemetery, Corringham, Essex, England, United Kingdom	5
8	53	Cemetery	1801	11747353	Corringham Cemetery, Corringham, Essex, England	4

Here's the Bath one (partial name search, first 100 shown), which is, frankly, absurd (4 farms in the first six entries?)

rank	score	displaytype	to-date	identifier	displayfullName	numJurisdictions
1	77	Town		12198468	Bath, Saint Thomas, Jamaica	2
2	77	Village		11047580	Bath, Saint John, Barbados	2
3	77	Farm		6067571	Bath, Orange Free State, South Africa	2
4	77	Farm		6067567	Bath, Transvaal, South Africa	2
5	77	Farm		6067565	Bath, Transvaal, South Africa	2
6	77	Farm		6067561	Bath, Transvaal, South Africa	2
7	77	Locality		3472358	Bath, Berbice, Guyana	2
8	77	Civil Registration District		1177831	Bath, Saint Thomas, Jamaica	2
9	77	Farm		673578	Bath, Upper Demerara-Berbice, Guyana	2
10	77	Farm		673564	Bath, Mahaica-Berbice, Guyana	2
11	77	County		394225	Bath, Kentucky, United States	2
12	77	County		393620	Bath, Virginia, United States	2
13	77	Village		12712089	Bath, Arwal, Bihar, India	3
14	77	Village		12690108	Bath, Bhagalpur, Bihar, India	3
15	77	Village		12678817	Bath, Darbhanga, Bihar, India	3
16	77	Village		12676295	Bath, Madhubani, Bihar, India	3
17	77	Hamlet		12062355	Bath, Freeborn, Minnesota, United States	3
18	77	Village		11943967	Bath, Tarn Taran, Punjab, India	3
19	77	Village		11932423	Bath, Bathinda, Punjab, India	3
20	77	Village		11910287	Bath, Amritsar, Punjab, India	3
21	77	Village		8124368	Bath, Carleton, New Brunswick, Canada	3
22	77	Hamlet		8046423	Bath, Park, Colorado, United States	3
23	77	Village		7205405	Bath, Lennox and Addington, Ontario, Canada	3
24	77	Locality		5845800	Bath, Cerro Gordo, Iowa, United States	3
25	77	Locality		5759059	Bath, Placer, California, United States	3
26	77	Village		5040102	Bath, Walker, Texas, United States	3
27	77	Populated Place		4937254	Bath, Northampton, Pennsylvania, British Colonial America	3
28	77	Village		4873360	Bath, Brown, South Dakota, United States	3
29	77	Neighborhood or suburb		4855123	Bath, Aiken, South Carolina, United States	3
30	77	Borough		4837903	Bath, Northampton, Pennsylvania, United States	3
31	77	Village		4826513	Bath, Summit, Ohio, United States	3
32	77	Hamlet		4822448	Bath, Allen, Ohio, United States	3
33	77	Village		4728492	Bath, Beaufort, North Carolina, United States	3
34	77	Town		4608935	Bath, Steuben, New York, United States	3
35	77	Town		4576371	Bath, Grafton, New Hampshire, United States	3
36	77	Hamlet		4194966	Bath, Knott, Kentucky, United States	3
37	77	Village		3991303	Bath, Mason, Illinois, United States	3
38	77	Neighborhood or suburb		3958700	Bath, Richmond, Georgia, United States	3
39	77	City		2242558	Bath, Somerset, England, United Kingdom	3
40	77	Populated Place		1914526	Bath, Saint John Figtree, Nevis, Saint Kitts and Nevis	3
41	77	Farm		705385	Bath, Wedza, Mashonaland East, Zimbabwe	3
42	77	Farm		705376	Bath, Gutu, Masvingo, Zimbabwe	3
43	77	Village		12230085	Bath, Phillaur, Jalandhar, Punjab, India	4
44	77	Village		12230084	Bath, Nakodar, Jalandhar, Punjab, India	4
45	77	Populated Place		8514361	Bath, Bath Township, Summit, Ohio, United States	5
46	77	Village		4651239	Bath, Bath, Steuben, New York, United States	5
47	77	Hamlet		4226242	Bath, Bath Township, Clinton, Michigan, United States	4
48	77	Village		4034808	Bath, Bath Township, Franklin, Indiana, United States	5
49	77	Civil Registration District	1996	12261098	Bath, Avon, England, United Kingdom	3
50	77	Poor Law Union	1930	3040256	Bath, Somerset, England, United Kingdom	3
51	77	Municipality	1877	1818873	Bath, Zeeland, Netherlands	2
52	77	Populated Place	1820	10551876	Bath, Bath, Lincoln, Massachusetts, United States	4
53	77	County	1776	3459031	Bath, North Carolina, British Colonial America	2
54	71	Populated Place		2382298	Bāth, Punjab, Pakistan	2
55	71	Populated Place		1326415	Bāth, Punjab, Pakistan	2
56	71	Populated Place		1326414	Bāth, Punjab, Pakistan	2
57	69	Borough		4837902	Bath, Northampton, Pennsylvania, British Colonial America	3
58	69	Farm	1980	6067568	Bath, Transvaal, South Africa	2
59	69	Farm	1980	6067566	Bath, Cape Province, South Africa	2
60	69	Farm	1980	6067562	Bath, Transvaal, South Africa	2
61	69	Farm	1980	705383	Bath, Wedza, Mashonaland South, Southern Rhodesia	3
62	69	Farm	1980	705372	Bath, Gutu, Gutu, Victoria, Southern Rhodesia	4
63	69	Civil Registration District	1974	3041448	Bath, Somerset, England, United Kingdom	3
64	69	Populated Place	1889	4873361	Bath, Brown, Dakota Territory, United States	3
65	69	Village	1867	7205404	Bath, Lennox and Addington, Canada West, British North America	3
66	69	Village	1860	9153553	Bath, Addington, Canada West, British North America	3
67	69	Village	1841	11854955	Bath, Menard, Illinois, United States	3
68	69	Village	1839	11854957	Bath, Sangamon, Illinois, United States	3
69	69	City	1801	10593849	Bath, Somerset, England	2
70	69	Town	1776	4728491	Bath, Beaufort, North Carolina, British Colonial America	3
71	69	Town	1776	4576370	Bath, Grafton, New Hampshire, British Colonial America	3
72	69	County	1705	3459030	Bath, Carolina, British Colonial America	2
73	67	County		415316	Bath County, North Carolina, British Colonial America	2
74	67	Township		4895294	Bath Township, Brown, South Dakota, United States	3
75	67	Township		4814001	Bath Township, Summit, Ohio, United States	3
76	67	Township		4813130	Bath Township, Greene, Ohio, United States	3
77	67	Township		4812552	Bath Township, Allen, Ohio, United States	3
78	67	Township		4717903	Bath Township, Beaufort, North Carolina, United States	3
79	67	Township		4343715	Bath Township, Freeborn, Minnesota, United States	3
80	67	Township		4107601	Bath Township, Franklin, Indiana, United States	3
81	67	Township		4074168	Bath Township, Cerro Gordo, Iowa, United States	3
82	67	Township		4036277	Bath Township, Mason, Illinois, United States	3
83	67	Village	1901	8015860	Bath-On-Hudson, North Greenbush, Rensselaer, New York, United States	4
84	63	City		4273070	Bath, Sagadahoc, Maine, United States	3
85	61	District	1849	3052420	Bathurst, Canada West, British North America	2
86	61	District	1849	3052419	Bathurst, Upper Canada, British North America	2
87	61	District	1849	3052418	Bathurst, Canada West, British North America	2
88	61	District	1849	3052417	Bathurst, Upper Canada, British North America	2
89	60	Village		4826514	Bath Center, Summit, Ohio, United States	3
90	60	Hundred		3039831	Bath Forum, Somerset, England, United Kingdom	3
91	60	Populated Place		8514362	Bath Center, Bath Township, Summit, Ohio, United States	5
92	59	Parish		2978934	Bath St Michael, Somerset, England, United Kingdom	3
93	59	Chapelry		2976573	Bath St James, Somerset, England, United Kingdom	3
94	57	Township		4717902	Bath Township, Beaufort, North Carolina, British Colonial America	3
95	57	Township	1889	4895295	Bath Township, Brown, Dakota Territory, United States	3
96	57	Village	1887	11932309	Possum Walk, Walker, Texas, United States	3
97	57	Parish	1801	10593865	Bath St Michael-within-the-Walls, Somerset, England	2
98	56	Town		5550394	Berkeley Springs, Morgan, West Virginia, United States	3
99	56	Town		5059602	Old Carolina, Walker, Texas, United States	3
100	56	City		4144198	Yates Center, Woodson, Kansas, United States	3

Rolf Ketil · May 3

From what is shown here in the comments, it seems to me, who is not a computer expert, but a genealogist with 50 years of experience, that the place name database should be restructured.

A place or address on Earth is fixed. This means that we simply use a latitude and longitude finder to find out where this is. This place that has been found is unique and must have focus, but what is interesting to us is that the place carries a history. Different people and families may have lived there, there may have been censuses, people were born, married or died there and the place may have changed its name at times. There are also local history books that are linked to this particular place.

In my opinion, a place should therefore be able to be linked to various variables, such as name, personal ID number, census and digitized books as well as other interesting material, photos, etc...

The Norwegian state via the Digital Archive has started a project where this is one of the goals. An English description of the project can be found here in a Historical Population Register (HBR).

https://home.nr.no/~holden/HBR-english.pdf

https://www.histreg.no/index.php/home

My recommendation is that FamilySearch seeks cooperation with this group so that a Historical Place Name Register can be developed as a "standard register" is not enough to find our relatives.

MandyShaw1 · May 3

I like this idea, but would suggest that the provision and maintenance of such a service would not necessarily be a specifically genealogical thing. Local historians and historical societies all over the world may have such information at their fingertips (not to mention people interested in the history of placenames, jurisdictions and elections, or historical geography). Is this about a service that coordinates the sharing of all this local information in a standardised way, and that can then be picked up by FS and other genealogical providers? I wonder if Wikipedia, or specifically Wikidata, may be a good source also.

But I think you are 100% correct that the fundamental problem with where we are right now is lack of context.

Rolf Ketil · May 3

Well, I get the feeling that we are on a more ideological track here, but are we back to square one where the place names should be in ascending order?

I have already had a dialogue with the management of the Norwegian Mapping Authority. And they have complained to me that their mandate has never been to log historical data, only to always be up to date. I think this probably also applies to other state institutions with the same task.

We can only hope that someone with sufficient resources will be called upon to gather the groups Mandy mention to put them together in a group where this becomes a topic that someone wants to tackle. In the meantime, I hope that whoever is responsible for standard place names will address the hopeless situation we are in now, which certainly needs to be improved.

Rolf Ketil · May 3

As a digression in relation to being optimized in relation to digitizing historical data, I would like to mention something positive for all of us.

National Librarian Aslak Sira Myhre and Archivist and Librarian of the Holy Roman Church H.E. Angelo Vincenzo Zani in audience with Pope Francis in November 2024 and here an agreement was made to be allowed to digitize the Pope's archive. The agreement can be found here:

https://www.nb.no/content/uploads/2024/12/Norvegia_avtale-1.pdf

FamilySearch has collaborated with Norwegian authorities in the past and should continue to do so.

Gordon Collett · May 4

To start with, let me say that I would certainly have no objection to reversing the time period order of place names. I have seen many examples where it appears users take the first name off the list and ignore the time periods. Reversing the order might help a bit with that.

I am also completely in favor of using correct historical names within reason and as generally practiced. For Norway I've seen that the majority of users of Family Tree and the majority of Norwegian users of Family Tree and My Heritage and even the Norwegian Archives fudge a bit with names. For example, in the archives all the historical records for what was Hordaland from 1919 to 2019 are listed under Hordaland. Nothing is listed as being in Vestland which it has been a part of since January 1, 2020, and nothing is listed for Søndre Bergenhus which it was called from 1763 to 1918 or Bergenhus which it was before 1763. It seems everyone has concluded that using Hordaland from pre-history to 2019 is fine.

Family Tree's dual entry of place names does make it easy to use Søndre Bergenhus and link it to the standard of Hordaland for people who want to use the older name.

The Norwegian Archives Historical Population Register is a pretty interesting simplified Family Tree type system in that it is designed to have one page for each person and to be a listing of all sources for that person. Here is the page for my wife's g-g-grandfather: https://histreg.no/index.php/person/pf01052269002108

It's like his Family Tree page in that it has his name, his birth and death information, a list of parents, spouses, and children and then a list of all the sources found for him so far with the information that has been indexed from those sources. (We've attached those five census records to his Family Tree and many more source from the archives that have not been linked to the HBR page yet.)

Regarding Place Names and the Places database, the biggest problem with it is that it is so far from complete. It was stated above that "A place or address on Earth is fixed. This means that we simply use a latitude and longitude finder to find out where this is. This place that has been found is unique and must have focus"

Exactly. And that latitude and longitude is the fundamental starting point of the Places Database. Each entry starts the that spot's latitude and longitude. To that is added a main name and as many variant names as can be found and may be useful. Then that base place is placed under each of its historical jurisdictions. At least that is what a fully completed single spot on earth will have. There are a lot of missing entries and a lot of only partially completed entries.

MandyShaw1 · May 4

I am more concerned about the type of place than about the dates issue, which is at least obvious from the list (if the database is correct in relation to dates, which is obvs a big assumption). I know we want it to be as granular as possible, and that even farms may have official records applicable to them only, but surely a setup which shows a farm in position 3, 66 places above a city whose applicability spanned from the 7th century to 1801, and 36 places above the 'current' entry for that city, has to be ridiculous? Filter options with sensible defaults are called for here, surely. Ideally in my view we would have the ability to filter the continent, too.

Gordon Collett · May 4

Filtering seems a bit much to expect of a simple dropdown menu. However, if you are working in the Places database to find the place you want to use, there are filters that seem to work pretty well from the little I have played around with them. This is getting a bit off topic, but I'll mention anyway that if I search in the places database for Tveit, I get about 180 results. But I can filter on place type, date range, and country (or any small portion of a county) to cut this down to 27: