Would this be a rule of thumb then? Always follow the newly discovered nodes and the ones on the opened list have a lower priority?
It looks like this particular algorithm's rule of thumb is to not even give those on the open list a priority. I'm guessing this is just for tutorial/demonstration's sake, but I would read his later points on how to improve upon this because this way, you'll have a hard time plotting a path through even simple mazes.
Can you possibly elaborate on "constant checking"? I'm a bit confused by the meaning of it.
If you scroll down to 8. Dijkstra's Algorithm, it's an example of an algorithm that does constant checking. Essentially, from the start, you check the F value of every node around you. Then for each node, you check the F value of nodes around those, and so on until you reach your goal.
Sometimes nodes will share adjacent new nodes. In the example, the node at 1S1E and 1S share an adjacent node at 2S. If you go from start to 1S1E(F=14), then to 2S(F=14+14), you get an F of 28, but if you go from start to 1S(F=10) to 2S (F=10+10), you get an F of 20. So initially 2S would have had a parent of 1S1E, but since it gets a smaller F from 1S (20 rather than 28), it changes its parent node to 1S.
Just that first paragraph should be enough to explain constant checking. This guy's example of non-constant checking is to choose the nearest from the closed list until he gets to the goal rather than considering the open-list, again probably having to do with demonstration, or explaining the basics
EDIT: didn't read the summary, yeah the guy considers the open list in the way I described