For me i consider average levels 6. Since you are curious about what the average for people is, I decided to say this:
Level rating judgement by BR:
1-2: Quite poor. Level needs severe fixing/remaking. Aka Crap.
2.1-3: Not pure crap, but needs severe bug fixing.
3.1-4: Level has at least 1 good point, but still needs lots of work.
4.1-5. Not that bad, but nothing spectacular and several things could be improved/fixed.
5.1-6. Around the average. Everything is decent, but nothing spectacular or interesting.
6.1-7. Ok, but could be improved and lacks some things.
7.1-8. Basic levels having good eyecandy, gameplay etc etc. May be some spectacular stuff.
8.1-9. Really good and a worthy download. Few cons.
9.1-10. Rarely levels reach this, but if they do it means they have very few/no cons and are amazing.
That is my standard. Feel free to comment negativly or positivly on it.
Here is a somewhat sketchy scale that I use:
1 – Horrible; no effort. Author displays absolutely no knowledge of JCS and has produced an unplayably flawed work.
2 – Some effort perhaps, but way, way too many mistakes. Author shows little ability to use JCS beyond the most simplistic of features, and often even messes those up.
3 – Little effort shown, but still a bit. Work is ugly and no where near finished. Author shows rudementary knowledge of JCS, but has not yet learned to use most of its functions.
4 – Some degree of effort is shown. Work could probably be salvaged, but not easily. Author shows developing knowledge, but no real understanding.
5 – Average. Work is neither good nor bad. Author shows a working, but superficial knowledge of JCS.
6 – Somewhat above average. Work demonstrates a good knowledge of JCS, but not a deep enough understanding or amount of effort to produce anything of lasting interest.
7 – Good, but not outstanding. The level looks good and plays well, but lacks anything worth of note. Author obviously knows how to use JCS, and well, but does not venture into deeper creative waters.
8 – Author has created a high-quality work. The work is interesting, at least partially original in ideas, and both looks and plays well. Level is, above all, enjoyable to play. Author shows a deep understanding of JCS and an ability to create very good works. However, the work is probably not likely to last for way too long and is not above and beyond.
9 – Author has created a work of exceptionally high quality. An expert knowledge of JCS is displayed, and the work is creative and shows an immense amount of effort in addition to skill. Author has gone above and beyond creating a long, playable experience which maintains high standards from start to finish.
10 – Author has created a work that more or less revolutionizes the field that it is created in. The author has a deep and profound understanding of JCS, level design, and the reaction of players, in addition to being a master. Legendary work that is sure to stay around for many years to come.
It just happens that the average level might deserve a six. So be it. I do not adjust my ratings to maintain a cumulative average. That would be a waste of time, really.
That is my two cents.