Largest unconnected branches size is underestimated

Question

Those who have already connected one of those might already know that : the largest unconnected branches are in fact larger in size than what appears at the dedicated FSP, based on the weekly WikiTree+ report.

WikiTree+ does not include private profiles, but if we add those, many of those branches are actually much larger.

It's possible to get the size of those branches using the Connection Finder query (used in the 100 Circles project).  For example the branch of Ramasamy-14 given in the FSP of size 311 is found by the CF query of size 2,810, almost a tenfold change!

I've started updating the FSP with such figures for the top 20 branches, but before going ahead and changing all the table, which would mean a drastic change indeed, what do interested people think?

A good side effect of using the CF query is also to find out in the table branches which are actually connected to each other through private profiles.

Comments

All answers so far here seem in favour of this change, but to proceed further I'm still waiting for a green light from the FSP managers.

---

I have written to Bernard privately about this, but I'll say it publicly here. I'm fine with him going ahead and revising the list to use the numbers from his search queries.

---

Thanks Greg for the green light

---

The table is now completely updated.

Answer 1

Wikitree hopes to be the one world tree for everyone and in that endeavor needs connections to be made. So I am all in favor or adding these larger private numbers even if we can't see who the private profiles are. As a long-time Wikitree Connector who uses that "Largest Unconnected Branches" page, I will be encouraged to get these larger numbers linked up, in order to facilitate even more world-wide connections in the future.

Comments

I would also like to see the CC7 numbers added to unlisted notables. Can this be accomplished as well?

Answer 2

I like the change! Is it also possible to get the number or the specific branches that exist on WikiTree+ for the unconnected branch? The true size is the most important, even if it won't be shown in WikiTree+.

I guess larger trees will get some more attention compared to the sum of the attention the smaller branches get.

Comments

Yes, we could add for each branch its WikiTree+ separate components. I'm in fact waiting for a feedback and green light from the page manager(s).

Answer 3

I would appreciate the size correction for large unconnected branches based on CF query over against FSP (WT+). Thanks, Bernard, for your excellent and innovative work!

Comments

Actually, in the current state of affairs, The CF query is not a tool to directly find unconnected branches. It just allows me correct the sizes found by WT+ reports.

And what I discover is that large branches can be hidden in WT+, split in several not-so-large branches connected only by private profiles. There might certainly be a way to program what I have started to make manually.

Go ask Aleš, I'm sure he'll know.

---

Ah, thanks for correcting my misconceptions! In that case, I’ll happily await the improvement of the apps, or rather the automation of your calculations …

More power to Aleš! ;-)

Answer 4

I need you to unpack what you mean by "Connection Finder query". I have never seen anything  about branch sizes in Connection Finder.

Comments

Hello Greg, let me "unpack", sorry for the abusive naming of the query. What is used is what we called in the early days of the 100 Circles project, the "magic query", which yields the distribution of circles population. This query is based on the Connection Finder data. It was intended to be used (sparingly, because it's heavy on the server - that's why it's not public) to get the data you can see in the 100 Circles and related pages.

But it works as well for unconnected branches, and that's what I used, e.g. for the largest branch Salsavilca-10

0,1,1,0
1,5,6,5
2,23,29,56
3,98,127,340
4,382,509,1384
5,1026,1535,6120
6,1651,3186,19445
7,1349,4535,39628
8,516,5051,52420
9,155,5206,55395
10,51,5257,56126
11,9,5266,56332
12,8,5274,56370
13,4,5278,56482
{"size":5278,"generations":14}

For each line you get : distance, number of profiles, cumulated number, number of connections traversed

And the summary : 5278 profiles in 14 generations.

The results include all profiles, whatever privacy level, like CC7 does.

The URL of the query can be shared privately.

---

And BTW the PM of most profiles in the largest unconnected branch, who herself belongs to the branch, Jenny Salsavilca Salsavilca, has a respectable CC7 of 3,759 (and counting), which did not help her so far to be connected ...

Answer 5

Hi Bernard,

Yes this would indicate to me and fellow connectors that there are more possible profiles to connect this branch. The branch-size that is reported at the moment, is only the part until the private profile and there should be a second/third branch for the part(s) "behind" that profile.

Answer 6

I'd love to dig into those private profiles. I'll bet you anything that there are profiles that should be open in those trees. I'm shocked to hear that there are almost 500 private profiles in the Gatewood tree. It would be great to get some of those opened.

Comments

https://www.wikitree.com/index.php?title=Special%3AMyConnections&w=Gatewood-910  shows indeed a lot of private profiles.
But the total number of those private profiles in the tree is not necessarily as high as the difference between the number given by the CF and the number of WT+ report.

I've pushed using WBE the above MyConnections page to display 1243 profiles, of which "only" 359 are Private.

WT+ does not follow connections through private profiles (if I understand correctly), hence some public profiles in the branch might be hidden for WT+ behind private profiles. That said you are probably right that more could be open.

Answer 7

Thanks, Bernard, for this very interesting discussion. I agree with the other posters about using the CF query to get a more accurate picture of the size of these unconnected branches. Statistics like these are real incentives to continue trying to make these more difficult connections.