I was aware of the slider at the top that controlled what degree of confidence and distance one can get from matches.

But all along I've had it set at around 60% figuring that would take in most of the usable matches. On a whim today I moved the slide out to around 90% and it looks like I gained several hundred new matches. Anybody have any insight into working far out on the edge like this? What degree do you have your slide set at, and has anyone gotten many usable matches out there in right field?

Also, I'm wondering whether it represents matches that are further away, or older, or maybe some ambiguous data in the DNA. Even before I did this most of my matches seemed to be out past the 8th generation or so anyway. Although I've had a few marked at 3rd cousin, I've never really found anybody quite that close.

Some of my best "hits" in terms of shared common ancestors have been found in the low to very low confidence 5-8 cousin matches.

The first one I tried gave me a solid hit at around the 23rd gr.gandf. level, around the 1100s.

Is it possible this test would pickup that far, or should I be looking for something closer in?

Well, your 23 g-grandfather might very well be an ancestor you share with your match, as well as with literally millions of other people. I can pretty much guarantee it is not your most recent common ancestor (MRCA) and the DNA test is not matching you because of that. You have over 8 million 23g grandparents, so there is no way to trace something in you to that particular person.

I may be wrong on this so people can correct me, but the DNA matching and the family tree matching are two separate things. Ancestry takes your DNA and compares it with the DNA of others, and when enough things they look at match they consider it a hit. When you click on that hit, they then compare your two trees, but that has nothing to do with what they found in the DNA. They know who you are in your tree, and they know who your match is in their tree, and they just simply walk up the tree looking for a common ancestor or ancestors by name. When you get a hit and you can't find a common ancestor, that just means neither of your trees has identified that person yet. It can also mean the match is incorrect. Think of the probabilities ancestry shows like the weather forecast. If they say 50% chance of rain, it may rain and it may not. The same goes for a 50% probability of a DNA match - it may be a match, and it may not.

I meant to add to that - this means there is value in the low probability hits. If there is a 20% chance that it is a match, then 1 in 5 of those is probably legit so it's worth your time to wade through them (depending of cours on your goals in all this).

I would say that 23rd ggp is too far back, even with pedigree collapse and recombination factored in... The lower the probability of the match, the smaller the segment is of shared dna. A low probability 5-8 cousin match could easily be a 11-12 cousin - especially with pedigree collapse. Personally, I have a good number of low/very low probability 5-8c matches that are really 10-12 cousins. Since the tree comparison tool only looks at the first 10 gens, I always look for something outside the 10 gen lists when looking at low/very low 5-8c matches.

Very interesting and thanks, McCorkleAustin. I have to say that all of my MRCAs have been 10 gen or higher. So, if I have a distant direct ancestor who is like Prince Halehumdra the Rattled, and I have a DNA match with a couple of other people who list Halehumdra in their tree, then it's probably just coincidence because Prince Halehumdra is listed in half the trees of English American immigrants, then the DNA match is useless?

If that is so I'm in a real quandary because almost none of my matches are within the first ten or so generations shown in matches when I first pull one up. So am I just confirming something more like we all come from Adam and nothing else specific on my tree? I do have a fair amount of people on my tree. But now I feel like I've really gotten nothing out of my DNA test.

Once you go back 6 or 7 generations, then yes, the DNA is not really telling you anything. If that person was indeed a common ancestor, you only have a teeny fraction of the parts of the DNA that ancestry is looking at, and the same applies to your cousin. The chance that you and your cousin have that same teeny fraction is pretty slim. Here are some numbers from the ISOGG wiki:

http://www.isogg.org/wiki/Autosomal_DNA_statistics

The numbers really drop once you get past 5th cousins. For example, on average you share only 0.00305% (that's 3 thousandths of a percent) with a seventh cousin. I would also bet that the information that leads you back to Halehumdra came from the same source that your distant cousin used.

The worth of the DNA project has to do with your goals. If you intended to use it to help with ancestors from the year 1100, then yes, you've wasted your money. Even though you can take a particular line back to that Prince or the 23g grandparent, I'm sure you have gaps or dead ends 4 or 5 generations back on some of your other lines, and DNA can indeed help with that. These cousin matches you are seeing 10 generations back could very well be from those much closer gaps. The point of moving the slider back is those matches could be closer than ancestry predicts.

I'm having massive mental agony over comprehending this and I don't consider myself a stupid person (although nobody really does, true of not). :)

I guess I really need to spend more time on theory, and shut up until I've done more reading.

It just strikes me that everybody wants to know about their grandparents and great grandparents. To a greater or lesser extent one might want to know about the people and happenings along his line of descent. But I, for one, couldn't care less about my 4th or fifth cousin living in Treekill, Montana, UNLESS it helps me nail down ggp 19, which it seems like you're telling me it won't.

Also, some people seem to be saying that a match for a 5th cousin, along with possibly being a huge meaningless mistake, might also show up in, say, a fourth generation, while others seem to be saying I should look higher, not lower, in 6th and 7th generation to find something.

Do the other types of DNA yield actual matches?

Ancestry's classifications of the closeness of the matches are based on probabilities, so yes, a cousin classified as a probable fifth cousin could be a fourth or sixth or seventh. You may have received smaller or larger matching segments than average just by chance. Another factor, I think, is multiple relationships. You may get a match if you are ninth or tenth cousins with a person three different ways. I may be wrong about this, but it seems true from my own experience. :) There is a post somewhere on these boards about why Ancestry's autosomal test may be more reliable than other autosomal tests with regard to the distant matches. If I can find it, I'll post a link.
Re: "Do the other types of DNA yield actual matches?" I am not sure that I know what you mean by this.