What is a fun level? I want my levels to be fun.

Know your audience. That's the first step in identifying what is fun or not. Even then it isn't 100% guaranteed though. Anyhow, how are the physics like? What's the age group? Art style? Mobile or different platform? iOS or android users? Are we talking about the animaniacs crowd for instance? Or are we making a game for Cthulhu fans?

So then, how could we make the game fun, with the physics. Borderlands art style with cartoonish explosions dropping from the sky blowing everything up, but with the right ragdoll physics it explodes and makes the bits and pieces from one creature kill the ones next to it. And then with a sound score that's some punk rock chick talking about chaos. There is probably a crowd that would find that fun.

So back to how to make a level fun? Know your audience.

Perhaps the 'levels' are not what's fun, but the 'gameplay.'
The levels offer you with more variation to keep you interested while you play through the game, but in a fundamental way they have the same gameplay* of the other levels.

Unless you have a minigame-centric game like Mario Party, your gameplay should remain practically the same throughout levels.
Game studios, even with their team of educated game designers, still rely on a Quality Assurance department with several workers playing the game and providing feedback to make sure that the game is not only functional but also fun.

So in order to have fun levels, you first need to make sure that you have a fun general gameplay to offer. This you can secure by designing the game in a way that you think is fun, and then by play-testing it thoroughly to confirm it.

This might be useful if you're studying level design: http://vgmaps.com/

* Gameplay as in, the result of goal-oriented player interaction with the game design and system.

Notice in Mario ( and to a lesser extent Sonic) that levels are usually broken up into a series of specific challenges. i.e.: a section with moving pipes and a gap in between. Avoid the flying enemies. Jump on the platforms moving in a circle. Avoid the crushing stones falling from the ceiling. Segment length varied considerably, but the segments themselves appeared over and over again in different levels with only minor variation.

So that's my recommendation--design your challenges in short segments, test each one, and build your levels out of these solid building blocks, rather than trying to do an entire level all at once.

Every once in a while a single idea is strong enough to do an entire level around, the forced scrolling levels for instance, but these should be used sparingly. Oddly enough a design is more cohesive with the series of mini challenges utilizing consistent mechanics rather than an overarching single challenge per level using a single unique mechanic per type.

I can't believe I forgot the #1 rule. Know your audience. This game is nothing special. It's 2D but isometric like the Zelda games but not quite. It an isometric platformer. There are no enemies so I'm kinda left with just well polished level design.

I always thought Rule #1 was make a games you like. Unfortunately, even knowing your audience doesn't equate to fun levels. There is no set formula for fun levels, but rather is just a matter of trial and error and seeing if it is a level you like. If you watch game dev videos, usually they say they made the game because it was an idea they liked and the levels, powers, etc. were things they did because they liked them and thought they were fun. Game devs take in player input and try to tweak the game for them, but ultimately it is things they think are fun. Others may disagree with me, but my mantra is "Make games you think are fun and like, and others will like them too." Not a 100% accurate mantra, but there is partial truth to it.

I recently played this game called Evac. I liked it but my game doesn't have the plot to make level like those. The game is really simple. All you do is move around but character talk to you and try to help you escape. Maybe I should have a character talk to the player and try to help them.

There are a lot of theory you'd need to apply to your levels, but it essentially boils down to something that feels right.

Playtesting is the cornerstone of good level design, but you still need to have a plan beforehand.

What you need to look at specifically:

- Pacing (does your game follow the Safe zone /Threat zone idea? if so, how do you distribute these safe areas, how do you reward the player with much deserved rest for overcoming many obstacles/threats?).

*Compare Contra with Super Mario 1 for example. You'll notice these are drastically different in terms of safe zones :)

*Don't forget to "break the pace" once in a while, to keep players off guard

- Distribution of gameplay ingredients (how often do you introduce new gameplay concepts to the player, and how do you revisit the same concepts in more complex scenarios?). A good example of this would be the first level of Super Mario:

*First, you are shown you can move in a safe haven.

*Then you are introduced to boxes, enemies and holes all in relatively independant segments of the game.

*Then, the level mixes and matches the above.

*The game slowly introduces variants: So you've seen a turtle? What about a turtle with wings?

You could have a perfect recipe of the above and still fail to deliver fun. You could totally forego the above and end up with something fun. However, out of 1000 games developed with either these methods or not, there will be more fun games that follow the above.

Key does remain playtesting though.