GENI does not return the most valuable connecting path

Tagging a profile with this problem as example:

Anna of Kiev, Queen Consort of the Franks

I show sensible / consistent path to her and her children, but Sergey flips to other lines. There have not been recent tree changes that we can see either to explain it, & as far as I know, no deliberate development changes in relationship calculations. Are we looking at a regression bug?

Private User any thoughts?

If you know the theory about Dijkstra's algorithm - Shortest Path in a grid, you would know that it needs quite heavy processing resources, and try to do it for 141,425,261 profiles... To solve this Geni re-uses already calculated paths and stitch them together, - not necessary the shortest path, and this algorithm has been better the last year where bloodline paths is getting a higher priority than a shorter path. Technically a path through a marriage is not a good path.

Anyhow, - I dont understand what you mean with "most valuable connecting path", - would you prefer royal and noble paths instead of peasants?

The same thing happens with me, when I have multiple lines of descent from an ancestor. Sometimes a known 24th g-grandparent will be reported as a 3rd cousin instead (for example). I suppose because the 3rd cousin relationship is through a "shorter" lineage path, I don't know.

Sometimes I can get it to show the direct line by monkeying with the program. I do it by dancing around the person whose direct line I want to see. Sometimes I can get the direct line to show up if I click on the spouse and then back on the one I'm attempting to study. Usually though it is just hit or miss for me, because I don't have many clues to help me figure it out.

And Private User it happens with any ancestor (not just the ones you might consider 'important' or 'popular'). In my opinion, ALL of my ancestors are equally important. Some are just more interesting to read about because they happened to have made history.

Private User I totally understand what you are talking about. I wish I could remember a line I was following where the female profile was my 23rd great grandmother and her husband (who was also my 23rd great grandfather) showed up as some sort of 11th cousin twice removed or some such. Just weird. If I find it again, I will post.

Susanne Floyd exactly. Something very similar happened to me just last night, but I didn't save it for future reference because it happens often enough to seem fairly commonplace to me, and I was busy checking out a large number of ancestors (which, as I'm sure you know, can be very time-consuming).

Maybe I should mention though, that the number of my ancestors has very recently dropped by about 2,500 +/- a few. And it's confusing not knowing how that happened.

A change in ancestors suggests a disconnect of an ancestral branch and yes, it’s confusing when it’s not obvious where this happened; if we could see the profile, we’d see the rationale. But that’s different from Dijkstra's algorithm, and recalculating relationship paths.

I’m guessing that Sergey has a “preferred” path to Anna of Kiev rather then her son.

"A change in ancestors suggests a disconnect of an ancestral branch and yes, it’s confusing when it’s not obvious where this happened"

Right.

"if we could see the profile, we’d see the rationale"

Not at all, because one has nothing to do with the other. It's two distinctly separate issues. One issue involves the "algorithm".

The other issue involves the lack of transparency with regard to tree cutting.

Some of that is not seeing the profile where the disconnect is.

As example, I disconnect profiles from parents that have been re attached despite clear notes saying parents unknown (we’ve only had relationship locking a couple of years). And won’t do a discussion because it was a tech error, or user error, not a genealogical dispute. And that could indeed impact “ancestor count.”

I don’t know much of a solution except for people to keep their own track of what they think important. Faustine was gracious enough to tag my Rev War ancestors (11!) in my profile “about”. Some members do their paths to interesting ancestors. And we have the HistoryLink application to make static reports.

This is where “teaching a path” is described:

https://www.geni.com/projects/Working-with-Relationships/17570

Erica, thank you for the helpful link. And thank you for letting us know that we are on our own with regard to understanding what happened to 2,500 ancestors virtually overnight. I'll probably figure it out, someday. :D

Erica Howton, that link on Working with Relationships is pretty good and supports what I have been finding - must click on every profile. One of the things I have started doing is checking in a profile to see if I am already following any siblings or children. It probably means that I am connected multiple ways since I rarely follow any but direct lines - though sometimes.

Yep, Faustine Darsey is gracious with her time, knowledge, and skill. She has helped me out on more than one occasion.

Thank you Susanne Floyd!

AND Thank you Erica Howton! I just saw your new profile photo. Cool!

I will clarify my last message a bit. I suggest to search first going by children of the next person down in path. The last children in path before myself will be my parents if the direct connection exists.

I posted the message below in my response to the curator. I decided to copy it also into this discussion.

I think that the problem is maybe deeper. The whole algorithm is strange (I decided not using stronger words). Like probably many people I put in my tree only direct ancestors ignoring siblings if the are not direct. Otherwise I would be flooded by remote relatives (have already direct 3,200+ in the tree). The only way to get the most valuable (direct) connection is to start the search from children of the person in question (like Anna in this case) going down via them into the newer times. The system will either reach me (reporting that Anna is my gr. grandmother) or will find that it cannot move deeper since all further children are NOT direct ancestors. Then it can report any first connection with indirect blood relation and any in-law connection if it finds some or none. It cannot be any other DIRECT connection expect the connection via children. In case of success the last child in the path will be one of my parents.
Instead they probably start the entire sort out of the tree branches and report whatever they find first.
Anna Yaroslavna in my tree has 6 connections: 2 to parents, 2 to siblings, 2 to children (both are my direct ancestors). In GENI she probably connected to more siblings. Anyway, in their connection of me to Anna they show her brother Vsevolod Yaroslavich (1030, my direct ancestors) as the previous person in the path and reported that Anna is my aunt. It means they started the search via siblings before they tried the children. Really bad! Of course,equally stupid is to chase the parents first since you cannot get DIRECT connection without including the children pointing to me.

Sergey Spivakovsky, I find this statement interesting:

"Like probably many people I put in my tree only direct ancestors ignoring siblings if the are not direct."

Two things strike me 1. These trees don't really belong to individuals and merges and additions take place in an effort to build the world tree. That is the purpose of Geni as I understand it. 2. I have some siblings that are my direct ancestors and it has been necessary to build quite a number of profiles to get to how they are connected to me. Some siblings have children that are connected to me in other ways. Example - The husband of my mother's sister turned out to be my father's fourth cousin. No one in the family knew. If a number of profiles and connections had not been made, this would not have been realized.

I get off on side-tracks with family members finding and building their trees and often find that they circle back to the family in another way due to their unique geographic location. This is where multiple pathways are a problem but in reality, they are a rich tapestry that is being woven as we speak.

Just for thought regarding not adding all the ancillary siblings - I might not know this:

Sergey Spivakovsky is Susanne's first cousin once removed's husband's second cousin 6 times removed's wife's 7th great nephew! (on my Mother's side)

Sergey Spivakovsky is your 20th cousin once removed. (on my Father's side)

Dear family member Susanne,

I should mention that I was talking about my private tree in MyHeritage where I copy the data from GENI. Yes, GENI is the world tree and the same Anna has there more siblings, and more children, and one more husband in addition to Henry I (my gr. grandfather).
Regarding the cousins, at this moment I have about 500 leads (matches from GENI and others) to more DIRECT ancestors. So, understanding that the task to bring all the cousins into my own tree is unachievable in the life time, I limit myself with those direct ones. Somehow I moved into the pool of people who are related to the whole Europe (Susanne gave one example) from Greece and Italy to Spain, England and Scandinavia. Some discoveries are really striking: I found that I related by blood to prophet Muhammad (via Russians->Germans->French->Spanish->Moores)! So, I put in the tree some interesting cousins, but not all of them.
Eating King Oscar sardines I decided to check relations to him (King Oscar II of Norway). And sure enough - we are blood related. Skipped him.

PS. MyHeritage owns GENI and reports all the matches from it.

Bjørn P. Brox, You are my 21st cousin 7 times removed. Answering about most valuable path, I ment DIRECT ancestors, not really nobilities. Still, it is a thrill to find that some historical figure is your relative. My grandchildren "ordered" me Al. Makedonsky and Julius Caesar :):). I got very close to them: Caesar's mother and his sister are my gr. grandmothers. And a close friend (from his childhood) and аlly of Makedonsky who substituted him in his absence as the head of Macedonia is my gr. grandfather.

Sergey Spivakovsky I know the connections between Geni and My Heritage and am a member of both.

I do not think it matters what tree you are speaking of really as the connections still manifest when you add siblings and multiple siblings who may be direct ancestors. Geni is going to have multiple siblings listed usually. I am not sure that It is fair to criticize a function of the world tree as it applies to your creating an individual tree. It is apples and oranges.

By the way, Geni reports all matches from My Heritage. I like Geni better as I can weed out the wrong data and have to manually add anything that is of use. Typically, I notify the profile manager if something is wrong, but sometimes it is so pervasive that it is just not feasible time-wise. I notified three MH profile managers last week about a death location that was impossible and likely an autofill. I work to properly source the Geni profile and get it mastered. Maybe it will filter back through over there with correct info.

I did know that I can modify GENI. Are you a manager-volunteer? They asked me to be a manager of Libin's branch where I added some data from Russian sources. And I submitted one serious GENI error in von Gerngross branch to the manager ( prooving it by our official family tree from Russia). In general I trust GENI and did not see many errors. Although there are little obvious things that I would fix like repeating the person names and town names twice and more. MyHeritage with its private trees contains a lot of unreliable data. People put in their trees whatever they want. I can make V.Lenin my grandfather. But their interface is superb comparing with GENI. I learned finally how to escape GENI's advertising asking to upgrade my service. I logged into GENI.com and then in other window use Google search: in GENI <name of the person>. Then clicking on proper geni.com in results I am in the profile.

What I find annoying while tracing ancestors is when it switches from maternal to paternal sides. I realize that coming from the same 2 counties in S. Jersey for just about 400 years and that my parents share, I believe 5 5 GGP's, it is still disconcerting!!!

Sergey Spivakovsky, if you mean "am I a curator or volunteer?" No. I just know that a lot of great people who are just normal everyday people like me do a lot of work on here and have helped me and others out tremendously. People like Emily Kent Marget and Cynthia Curtis, A183502, US7875087 and Private User and Dr. R. Owen Wyant, (PhD) and the list goes on have helped me tremendously while working on their own projects. We are all connected somehow and it is nice.

I would not know about all these folks if paths did not get tangled. The curators are in a class beyond and I appreciate what they do more every day even when it doesn't fit my story. As I have said before, try as I might, Uncle Elbert was wrong. We did not descend from Pocahontas. Maybe another noble native, but not Pocahontas that I can find and so it isn't going on the tree. :-)

Private User, I feel your pain. I have the same thing happening with my Horry County and Marion County, SC relatives. Been there since dirt.

:)

Thank you! I try my best to give the most accurate infromation always.

Susanne Floyd - Thanks for the shout out.

Does anyone know how close GENI is integrated with MyHeritage? If GENI relies on my tree in MyHeritage where each person has a tag DIRECT ancestor (or NOT DIRECT in a very few cases) then the search starting from children would be very fast. Search in GENI does not know about my connections and run the search for each person. Still, it was a very quick correct return when I asked about connections to Elizabeth, Countess of Leicester, Surrey (born De Vermandois), my gr.grandmother, and daughter of Hugues I 'Magnus', Count of Vermandois..But GENI produces wrong result (first cousin 25 times removed) after relatively quick search.of Hugues. The only explanation of that is that they run incomplete search returning the first path found. They obviously don't go to my tree. Nothing in GENI was removed or changed in the branch connecting me to Hugues via his daughter. GENI also seems ignoring the results of the previous search. When I found my DIRECT path to Elizabeth and in her profile clicked Hugues they still gave their wrong answer although it needed just a single step if they keep temporarily my previous search.