Combining filters

[Eirikst]

Hey,
There is this thing that got me confused regarding combining filters. This is my impression:

I have a network where all nodes belong to one of three categories (countries in my case). In the initial face of analysis I run stats such as average degree. Now I want to filter out all nodes that does not have edges across the categories (countries). I then add a inter edge filter and tick all the boxes (the countries). This removes edges that does not cross a border.
This does not remove any nodes. So I combine it with a range filter and use degree in range 1-max, in order to get rid of the once with 0 degree.
I learn from this tutorial (http://www.clementlevallois.net/gephi/t ... ons_en.pdf) that the inner most filter is run first, so I must run the range filter first, and then use the inter edge as a sub-filter. When I know this, it somehow makes sense.

I press the filter button, but the expected result does not happen. The context box updates with new and lower number of visible nodes and edges, but not a brutally as I expect.

Then I re-run the average degree from the stats panel, and the run my filter(s) again, and the expected result appears. So far, I get the results I think is the correct, so good. Right?

But then. I want to tweak the range slider a bit, see how many nodes that disappear is the threshold is higher. I try with range 10-max. Edges disappear again. Not nodes. I run the stats again to get updated degree numbers. I run the filter again and I get what I look for (I think), nodes are removed. But then I go down again. Range 1-max - and expect to get the same as the first time I had 1-max range. But I dont. I stay with the graph I had at 10-max. So I rerun the stats again, and rerun the filter again, and the I get back to the first 1-max results.

Is this just me or is this super confusing? Is it the intended way for this to work? The chaining of filter that involves some part of the filter to interact with stats than take part of the next element of the chain should, at least in my opinion, automatically rerun the stats that are involved, somehow. It is confusing to figure out at what stage of the process I'm in, and what actions that must be done to complete the filtering. I notice that results I have noted down from previous sessions of analysis are wrong, since they are written down without some stats re-run.

I cannot find any posts involving chaining filter here, but am I doing this right? Is this something you have run into and figured out a good process for? Is is meant to be this way?