Sources are Disappearing Left, Right, and Center! What's going On?

LegacyUser

Jeff Wiseman said: Something serious is going on. I saw a couple of these earlier, but now I'm seeing dozens of them!

Source citations attached to and referenced by (tagged) conclusions in person records which have been closely researched, vetted, attached, and internally referenced are being corrupted in droves as the URLs in them are failing. Here's an example:



This citation only contains the URL now. Trying to access it results in:



It used to contain the following content:

"Ohio, County Marriages, 1789-2013," database with images, FamilySearch (https://familysearch.org/ark:/61903/1...), Pearl Harrison and Alice Hawkes, 12 Jan 1918; citing Marriage, Pickaway, Ohio, United States, pg 203 n 405, Franklin County Genealogical & Historical Society, Columbus; FHL microfilm.

Which had far more information in it. Furthermore, in this specific case, it was the ONLY source that not only tied the husband and wife together, but also tied him to his parents. This sources was referenced in several places and now it has died making all of those notes and "Reason this information is correct" notes useless as the referenced source can no longer be accessed.

And many of the person records I'm working with have a half dozen broken sources like these attached to them.

So I no longer have dozens of sources which were referenced from all through the person records that have now disappeared, nullifying any benefit that may have come from referencing them in the first place!

What is going on? Is this just a big mess up by FS that will be fixed and the records returned? Or are these records all permanently gone, eliminating all the usefulness of ever spending all the time to attaching them as sources in the first place? Am I going to have to now do repeated searches to try and replace all this information again?

I've already found where folks are going through and detaching all of these because they are broken. And yet many of them have no other sources to replace them with. The original sources were adequate, the images readable, and the citations legitimate. What is being done to correct this loss (if anything)?

And I am getting no change reports on ANY of these deletions! Assuming it stops now, it is still gonna take me weeks to replace all of these (assuming I can find where they have disappeared from)!

LegacyUser

Brian Jensen said: Jeff,
Engineers are aware of this issue and are working to resolve the issue.
Thanks for your patience.

LegacyUser

Jeff Wiseman said: Thanks for the response Brian!

Do you have any suggestions for the time being? I assume that detaching these broken citations from person records is not a good idea at this time...

I also assume that dismissing hints that are broken this way probably should not be done either?

This looks a little scary based on the potential rework I may have to do on dozens of records.

LegacyUser

Paul said: Jeff

Can't find the thread just now, but this was reported some time ago. I assume the engineers are working hard to fix it, but it is a potentially damaging issue. The main problem will be, as you suggest, if users think these sources are now permanently worthless, so decided to detach them.

I have found it unwise ever to detach a source, even if it has been redundant for some time. Some years back, a whole new set of URLs were introduced for the 1851 England & Wales census collection. I edited the titles with the suffixes "New version" and "Old version". However, the "new version" contained far less detail and, after a while, the original URLs reappeared! I now have to remember that everything I had marked as "Old version" is the source I should refer to - and obviously to ignore the "New version" sources! I dare not detach the latter, however, as who knows that when the collection is next updated it might be THOSE URLs that are used in the update!

In fact, I dread each (E&W census) update coming across from Find My Past, as there has never been consistency in whether the URLs remain the same for the matching records or if a new set of URLs will appear. I don't know what will happen will the current "410 error" problem with some of the (I believe) 1861 E&W collection, but hopefully I will be able to reaccess them (in whatever form!) fairly soon.

Hope your specific "410" problems will be resolved soon, too.

(Update - just found topic I raised 6 days ago at https://getsatisfaction.com/familysea...)

LegacyUser

Jeff Wiseman said: 6 DAYS? Wow. FS has the ability to put up notifications for when parts of the system may not be functioning. Could they at least have put up a notification that people currently "should not detach sources or dismiss hints that have 410 errors" in order to mitigate the loss of previous work done on with sources on person records?

I see people currently going through and detaching all kinds of these sources that I did work on years ago.

LegacyUser

Paul said: This is a (part) copy/paste of my comments at https://getsatisfaction.com/familysea.... PLEASE could a FamilySearch employee explain what really is going on, before this matter causes some damage?

I have just been investigating some reinstated sources and believe I have discovered a likely reason this problem is occurring.

It appears some comparison in "duplicate" sources is being undertaken by a FamilySearch team and this has led to many apparent duplicates being removed. The problem is, they seem to have got it wrong and have been retiring records that are not duplicates at all!

Some time ago, Brett complained about the removal of 1881 England & Wales census sources. He has commented (on another forum) these have largely been reinstated. Ironically, the detail these sources (dated 2017 & 2018) contain is pretty much the same - only the detail is listed in a different order. Unlike Brett, I feel one set of these could have been made redundant without any loss to users. However, many of the sources that remain withdrawn (showing "410 - Record removed") do NOT seem to duplicates of other sources in any way, as I say.

I have argued this issue (particularly with Brett) over a long period. I now have to admit he appears to have been right about the suggestion by Robert Kehrer (a few years back) that an exercise would take place to remove the "least valuable" of these almost identical sources. I thought this was a great idea. However, if my theory is correct, then sorry Brett, it appears pulling sources in this way can prove to be a dangerous exercise!

Of course, I could be completely wrong and other reasons might be involved. Sadly, conjecture will continue to be rife until a FamilySearch employee is willing to share the facts with us, concerning why so many of these sources (from a wide variety of collections) are currently being "removed".

LegacyUser

Paul said: Regarding the 1881 E&W census sources that apparently were marked as "401 - Record removed" for a while. The page at https://www.familysearch.org/tree/per... shows I have added one source for this today - it is dated 21 February 2018. The other was added 22 August 2017, but is dated 11 December 2017.

The problem with these E&W census collections (which are provided by Find My Past) is that sometimes an update "overwrites" the existing URL(s) but other times they appear (as an update) with a different URL. In the former case, there is just one source available, but in the latter there are (now) two. However, as in this case, only one source will be found in a search at https://www.familysearch.org/search/. I assume the one I found today is for the reinstated URL, as I cannot find any indication as to which of the two sources it relates - see https://www.familysearch.org/search/r..., first result.

The detail given in both sources appears much the same, albeit the details are shown in a different order. The oddity of the date one of the sources added to this ID being nearly 4 months BEFORE the date of the source date probably relates to the anomalies involved in these sources coming over from FMP, as explained.

I'm sure users like Brett will be delighted to see these sources reappearing, but if the more common procedure (of these updates coming across from FMP) had been applied, there would never have been two sources / URLs appearing at all, in the first place!

A point I was trying to make in my previous post is that these extremely close matching sources would not really be missed if one of them was retired (as, usually, one wouldn't even be available). However, the serious problem is if two completely different sets of sources are being treated as duplicates and one being removed in the belief (of someone at FamilySearch) that it is a true duplicate of another.

But, as I have said, maybe I am off-target here - as there doesn't even seem to be a similar source available (to the one being removed) in some of these cases.

LegacyUser

Paul said: Oh dear, apologies to Brett again! Have just checked 2017 and 2018 versions of 1881 E&W census (see above) and discovered the 2018 one DOES have better detail: house number, as well as just street name, included and extra members of household are also shown. This doesn't alter the fact that only one set would usually be available on this website.

A search at https://www.familysearch.org/search/r... shows only one set can be accessed from there. Unfortunately, it is the 2017 (earlier) version, which contains the lesser detail!

So, fine if you have been using FS for some time, but not so good if you are a more recent user.

See https://www.familysearch.org/tree/per... and above link (Benjamin Tungate born 1825 Upton & family in 1881 census)

I know this appears to be going off-topic, but this clearly shows we do need any "410" withdrawn sources reinstated, IF possible, as SOON as possible.

LegacyUser

Mike Lynsky said: The fix has deployed -please verify

LegacyUser

Mike Lynsky said: This has fixed -please try again

LegacyUser

Jeff Wiseman said: So I just had a quick check on one of my ancestors and found 4 bad citations. This are all on Edward Wayne O'Bryant KVGC-47D, and the bad citations have the following titles:

- Edward Bryant in entry for Cutler O. Bryant and Nettie Schiller, "Ohio, County Marriages, 1789-2013"

- Edward O'Bryant in entry for James O'Bryant and Minnie Dearth, "Ohio, County Marriages, 1789-2013"

- Edward O'Brien in entry for Harley O'Brien and Mary E. Freeman, "Ohio, County Marriages, 1789-2013"

- Edward O'Bryant in entry for Wayne D. O'Bryant and Cecil F. Santee, "Ohio, County Marriages, 1789-2013"

1. However, The problem on these has seemed to have morphed into something a bit different now. If you open the source in the source list, it has the 410 - Record Removed Error. However in all 4 cases, the "ark" style URL given under the URL label is DIFFERENT from the "ark" style URL under the Citation label (this is also the same under the View and Edit tabs):



2. And in spite of the 410 error displayed, selecting either the URL under the URL label, or using the (different) URL under the Citation label will actually take you to the same source in both cases (each with its own URL value).

3. Also, for some reason, the citations being created are now missing a LOT of information that was there before (such as film numbers, page numbers, etc.). These are the types of things that are supposed to be in a citation, but they seem to no longer be created in citations to indexed sources. Here's a example of an earlier citation copied from FS to my AQ database (left hand side), to the current FS citation (on the right hand side):



4. And one last item. Although this might not be directly related to the removed sources problem, there are multiple alternate date values that are showing up which are obviously totally wrong:



Note that I have only looked at just one PID at this time.

LegacyUser

Jeff Wiseman said: Thanks for reporting back to us! See my problem report below.

LegacyUser

Jeff Wiseman said: one more note. I just went back to review my notes and it appears that the fist of the four sources I referenced (I.e., the one I referenced in all the images attached), no longer shows the error and will display the correct page.

However, the other 3 issues still exist.

LegacyUser

Tom Huber said: Wow. Ohio? I have at least two of my ancestral lines who largely lived and still live in Ohio.

LegacyUser

Jeff Wiseman said: Yea, pretty well all of mine are centered around south and south-eastern Ohio. Nobody seems to have been farther north than Columbus. My great grandmother was a Ross. Ross county (where I was born) was originally settled by that family.

LegacyUser

Tom Huber said: One of my ancestral lines were in Brown, Adams, and Scioto Counties where many descendants still live. The other was in Coshocton County where many descendants still live.

LegacyUser

Jeff Wiseman said: i'm not so sure about the Coshocton connection, but yes, Brown, Adams, and Scioto counties fit the pattern. When Ohio was being settled, they came in either form the lakes and Erie canal in the north, or up through Virginia (now Kentucky) on the south. They crossed the ohio river, and my ancestors then migrated north and north-east mainly driven by the evolving iron-ore industry in those areas prior to the civil war. Looks like yours spread out along the river.

LegacyUser

Adrian Bruce said: Re the absurd birth dates - I had a look at the first image that Jeff highlighted and paged onwards several screens - all are infected with utterly absurd birth dates.

What's happened is that the original documents in at least that part of Ohio, County Marriages, 1789-2013 give the person's last(?) birthday consisting of dd mmm and yyyy (yes, in that order!) The yyyy is used to calculate the year-of-birth to go onto the index / Historical Record. The mmm dd values (in that order note) have gone into the birth date item in the index / Historical Record, where they have been subsequently interpreted as if they were mmm yyyy!!

Since the so-called yyyy is only two digits, some sort of estimation (well, guesswork) algorithm has been used to come up with an array of possible four-digit-years.

So for Cutler Bryant, what was 29 April on the document has clearly become April 0029, which is interpreted as a 2-digit year and the algorithm has decided is probably April 1929 (for a 1905 event!!!!), but possibly April 1829 and, just for form's sake, April 0029, i.e. April 29 AD.

I have no idea how far that nonsense goes but at least some of that Ohio, County Marriages, 1789-2013 collection needs to have its index rebuilt to interpret the dates correctly. Can't blame the poor indexers!

LegacyUser

Jeff Wiseman said: For some additional side effects, see:
https://getsatisfaction.com/familysea...

LegacyUser

Mike Lynsky said: Although we have fixed some of these data problems, there are some remaining issues we still need to address. We take data loss very seriously and because of your input, we believe we have identified where this happening and why. The resolution requires work across several teams which are being coordinate now. As soon as we believe we have the fix, we'll post back to the several threads discussing this problem and ask you to verify if you are now seeing the correct data.

Thanks for your patience

Mike Lynsky

LegacyUser

Tom Huber said: Thank you for your report, Mike. It is appreciated.

LegacyUser

JimGreene said: Is there any lingering issue with this now? Can we assume they fixed it to everyone's satisfaction? If not, I'm all ears

LegacyUser

Jeff Wiseman said: Jim Greene,

The original problem with the "401 Record not found" issues seems to have gone away. At least, I haven't seen any for a while. However, I am still seeing issues related to all 4 of the additional items I listed above showing up, and one additional one now:

#1 & #2: In spite of the fact that the 401 error has gone away, the mismatched URLs are still showing up and resulting in true duplicates of citations. I'm pretty sure that #1 & #2 are better covered already in the topic:
https://getsatisfaction.com/familysea...

#3: I am continuing to see this. The citations being created from hints and when sources are attached via the source linker are missing almost all of the rich citation information that used to be inserted into them. Note that this may have been planned but you now don't get much information at all about the source in the citation. If you can't access the source, the citation can no longer help you.

#4: The random nonsensical indexed dates seem to keep showing up in different citations. Not sure why but they are still their in many records that i see.

#5: This is new but related to the original problem There are now several sources that have been only partially "attached" to a person's Source List. These appear to be source citations that were damaged when people attempted to remove them when they had 401 errors. These can be found with the unfinished attachment feature. For example, Edward Wayne O'Bryant KVGC-47D currently has 5 sources in his source list that have are not attached to ANYTHING.

Also from the following image you can see that these sources had been attached years ago:



Opening the 1910 Ohio County Marriages source for Edward shows how even though there is a citation to this source in Edward's source list, it is not attached to him (or anyone else for that matter):



And finally, if you examine the contents of that same citation, you get:



As you can see, these have the same mismatched ark URLs as I described in #1 & #2 above and in:
https://getsatisfaction.com/familysea...

Although the citation was created in 11 Nov 2015 when I first attached the source to everyone, around 2 months or so ago, the citation text was partially modified in the existing citation to show a new ZZRQ-N1W2 ID.

So I have to go through all of my ancestors manually now to find every source that has been totally detached from all of the family members that they were originally attached to years ago, and yet still is showing in the source lists for them.

LegacyUser

Barbara Nelson said: I am still getting 410 record removed for the 1870 Houston, Minnesota census.

LegacyUser

Mike Lynsky said: Barbra -can you send me a screen capture of the problem you are seeing and we'll look into it

LegacyUser

Barbara Nelson said: Mike, I'm sorry, but it is not letting me add the picture.

I did a search for Peterson, born Minnesota 1860 to 1870, and match residence exactly of sheldon, houston, minn

I selected Eliza Peterson, born 1867.

It than shows a result for Halvor Peterson, PID GWCF-CPB​​ with a "! 410 Record Removed Error".

Hope this helps,
Barbara

LegacyUser

Mike Lynsky said: Thanks Barbra -how about copy and past the URl from your browser for the problematic record -it should take me to record your seeing the 410 error -thanks

LegacyUser

Barbara Nelson said: https://www.familysearch.org/ark:/619...

Here it it. But that doesn't show what I see as I see what I stated above.

LegacyUser

Mike Lynsky said: Sorry for the slow response -pls try the url again and see if it's fixed. We're investigating if there are other personas in that collection that are broken

LegacyUser

Barbara Nelson said: I am no longer getting a '410' when I do a search for Eliza, but I should be taken to this page of the census.

https://www.familysearch.org/ark:/619...

The search shows this, which links to a Yucatan, Houston, MN page.

https://www.familysearch.org/ark:/619...

I did a search for Halvor Peterson also of Sheldon, Houston MN and get a 410 error.

https://www.familysearch.org/ark:/619...

Somehow Sheldon Twp and Yucatan Twp have gotten confused.

LegacyUser

m said: Not good.