Sorry, I forgot this.
My idea is the following.
There's two main points that this suggestion tries to allow, in addition
to what has been proposed
a) We would not want other systems just to be defined as "below" the
railway one. For example, I'd like to see an important underground
interchange station earlier  than some unimportant regional train
stop; and some small underground station should be shown later than an
important bus node.
b) We need rather identical schemes for several systems: Airports,
long distance trains, long distance buses, regional trains, regional
buses, suburban rail systems, fast urban rail systems (underground),
slow urban rail systems (tram), (sub)urban bus systems. Yes, some of
them lie outside the scope of openrailmap, but there's the
same problem when rendering a transit map, so we should develop a
solution that fits them all.
Thus, my suggestion:
For each of these "levels", there are four "importances" values:
- important node (n+)
- secondary node (n-)
- important stop (s+)
- secondary stop (s-)
So for every station, we would store how important it is within every
"level" of traffic it is served by. 
We'd have to think of nice words for the importance values, but for
demonstration I'll use the n/s +/- codes I noted above.
importance:longdistance = s+
importance:regional = n-
importance:suburban = n+
For the "long distance" level, we'd have:
Frankfurt: Hbf (n+), Süd (s-), Airport (s+)
Berlin Hbf (n+), Gesundbrunnen (s+), Wannsee (s-)
Mannheim (n-), Heidelberg (s+), Darmstadt (s-)
München (n+), Nürnberg (n-)
Köln: Hbf (n+), Messe/Deutz (s+)
Bielefeld (s+), Gütersloh (s-)
Freiburg (s+), Offenburg (s-)
For "regional traffic", we'd have:
Berlin Hbf (n+), Ostbahnhof (s+), Potsdamer Platz (s-), Südkreuz (s+),
(I don't know other regions well enough)
- because only the importance within one level has to be considered,
only four (maybe even three) importance values will do, and it is quite
easy to define which one fits.
- maps can decide how to rank the importance values of the different
levels. For example, I'd say an important suburban traffic node is to be
ranked higher than a secondary regional traffic stop. For example,
Berlin Westkreuz is an important S-Bahn node, while Jungfernheide is a
secondary regional traffic stop, thus Jungfernheide can be displayed
later than Westkreuz.
- We'd have to define and store several values for most stations.
- And these are mostly redundant: Of course an important long-distance
stop is also an important bus stop, etc.
So maybe this level thing can be "compressed" into one thing, maybe not.
Let's first discuss if it is a good model the way I suggested it.
PS: For me it's a lot more work to write this down in English in a way
that it hopefully can be understood. Are there even non-germans on this
list? (German as the language, not the country :) )
 earlier meaning "appears on a lower zoom level"
 considering OpenRailwayMap, we have the additional problem that one
station can be several ones technically. For example, this is the case
at Berlin Hbf. Here we would want to first just display "Berlin Hbf",
but later (when zooming further in) display labels for "Berlin Hbf
(tief)", and the S-Bahn and U-Bahn stations.