OK, if you look in my family tree for William James Govier (1854-1915), one of his children, Henry Govier, is down as having been born "Between 1884 and 1887". The online family tree in both classic and modern mode (which really seem like they're the wrong way round to be honest, but that's another issue) can't make sense of this kind of inexact date. In classic mode no date is shown unless he is selected, in modern mode the right date range does appear. However in both modes it forgets that he should be the 6th child and moves him to the end, making him the last child.
The same issue can be seen in my family tree when only one sibling among many has a known date of birth. Lucia Recinelli should be the second child of seven after her sister Rosaria, and is specified as such in Family Tree Builder, but in the online family tree in both modes she appears as the first. When dates of birth are not known it seems to prioritise male children first as it has placed Alessandro second online when he should be third.
I'm glad to tell you that the first issue that you presented with the children of William James Govier, we fixed. We now consider the earliest date when specifying "Between 1884 and 1887" in the new family tree display (which is not yet available to the public).
Regarding the second issue, if the dates of birth is unknown then the family site would not know how to prioritize. We would recommend you to do as you've done in the first issue to specify a range of dates and organize them by the ranges.
I hope it makes sense, when the new display will come out you will have much less consistency troubles with the family tree.
Good one, Noam; not exactly what I had in mind, but good one nonetheless ;)
Here's the deal: "There are 2,121 Smart Matches™ you confirmed" says MH. You bet I didn't check each and everyone to see what info I could drag over to my tree. I did check them overall to see if they were talking about the same people, and duly rejected the ones that weren't. Now, what'd be great to have is MH to say: "Hey, you've already confirmed matches for these people (and so have their tree managers), but what do you want to do about all this info that you don't have or doesn't match with your tree? Do you want to copy it over or leave it as it is?"
What would be also great is for MH to tell me the info on the people I matched has changed in other trees, so I can choose to update mine. Theeeen you'll see some consensus building across your family forest ;)
Noam, in sorting by value add, it'd also be interesting to consider the distance of the matched relatives to the tree owners. I sort of already do that manually when comparing the facts.
OTOH, don't think simply resorting cuts it. In fact, I'm fairly happy with the confidence sorting as a proxy. What I was looking for is a screen that lets me copy me details across for many individuals in bulk rather than one by one. It'd highlight the facts of all individuals in my tree that are different from the consensus or from the tree of their closest relative. I would then select all that I want to copy over to my tree. That'd be much more efficient.
You are right that it would make sense to push forward matches that reveal new information about close family, which will often be more interesting to the tree owner.
As for allowing the user to do bulk merges, we are vary careful about this.
We want to allow our users to add new information to their tree from matches selectively and one by one, but don't want to enable harvesting of information. The reason is that many genealogists are very sensitive about their trees, and we want to find the balance here between cooperation and privacy.
What we are working on (and will release very soon) is a Save Record wizard, which will highlight new facts from Record Matches and will enable you to merge them to your tree. We already have something similar in Smart Matches but this will be more advanced. If we see that it is a success we might consider adding it to Smart Matches as well.
I understand the balance issue, Noam, but someone that really wanted to harvest information could easily build a script (in, say, Greasemonkey) to copy all the info in bulk, one by one. You probably need to think of other safeguards that will protect privacy but not at the expense of accuracy.
On the point of closer family, I actually meant the exact opposite: I'm more interested in learning about my relatives that are closer to other tree owners than on what info they have about my close family (for which I am more likely to have the best sources).
That needs to be defined, as each item on the list of "by people" contains many smart matches, some of which might have high quality and others low. I am not sure that consolidating the quality would mean much here.
On the other hand, we are planning to do just that with the added value sort. We will consolidate the added value in all the matches for a specific person, and present the final score to you. You will then be able to sort by added value in the "by people" page, and see the individuals whose matches will bring the most amount of new data to your tree, first.