I guess I should post my opinions here

1) It is true that everyone has their unique style, but that doesn't mean that all styles are equal in quality. In the end of the saw you are drawing something, and as with anything you draw the better it looks the better it will be rated. While it is true that some styles do look better than others, someone with good drawing abilities would make somehting better using a certain style that someone with bad drawing abilities could do.

While the drawing quality of a set is a big factor of it's overall rating, it is not the only one. Tilesets are not only made to look good but they are also made to create levels with. If a user is unable to create a decent level with a tileset or if the user finds that the tileset limits his levelmaking abilities (eg. lack of needed tiles, difficult to understand tileset layout or a lack of usability with the tiles given) then that user would obviously rate it lower regardless of how good it looks. Note that the above example assumes a general level of skill of the levelmaker and would not always be the case in the real world. A good levelmaker could still make a good level with a bad tileset, while a bad levelmaker could still make a bad level with a good tileset. The point is that the good levelmaker would be able to make a better level with a good tileset than with a bad tileset and vice versa.

2) "Sets' styles and differences are what that makes levels vary, tile quality and quantity means a lot when creating a level."
I agree 100% with that

"For example, a set such as Swamps by Agama would not suit very well to a simply and flat level as it's far too detailed, while a level with characteristic depths would probably require it."
However, I don't 100% agree with that.

The amount of detail in a tileset does not determine it's usability. There are a few cases where this is not true, which is at the very extreme when a user sacrifices tile usability for increased detail (eg. making tileable areas 2x2 tiles instead of 1x1 to increase the variation in the tiles). Mez01 is still a detailed tileset and has a lot of usability. While swamps is a detailed, it doesn't have the usability that mez01 has, but in my opinion this is not due to detail. In my opinion the usability of a tileset is determined by the masked image it usage. After all, the mask determines how the character interacts with the tileset, not the way the actual tileset looks. If the mask is done correctly, it will fit perfectly into the detail of the tileset and there will be no, or very little loss of usability for a great gain in detail.

You also slightly mentioned a bit about themes, but didn't go into detail about them. I do feel that it is an important topic and will expand on it a bit more.

It is true that some themes are more populat than others. In general there are three reasons for this:

a) The theme is easier to make in terms of detail. A few bricks and a road are easier to mak than textured sand and individual grass blades or the like. You could still make a very detailed city-themed set without using shapes that are difficult to draw.

b) The theme is more popular in terms of use. Face it, people like some things more than others. For most people the chance of someone actually using their tileset is low...very low. Why risk having your tileset not used and make something you know not a people like if you could make a space or jungle set instead that you know people use often?

c) The theme has more things to do with or add detail to. Compare a desert to a jungle. In a desert there are very few objects to work with, because a desert is just...well...empty. It is true that with creativity and some imagination you could come up with things like cacti or an oasis, but overall a jungle has a lot more to work with and would be a much easier choice to add detail to. It is mainly because of this reason that I tend to use some themes more than others.

3) I am not quite sure what you mean in the first paragraph even though you mentioned me, but I will try and comment on the pieces that I can interpret.

First and foremost, you are correct that creating more tilesets will improve your skills in tileset making. This is not the case when you keep reusing tiles, as the actual experience you get is when you go down to the nitty-gritty and drawing things over again. Even if the result of the second or third attempt doesn't look as good as the first, you could atleast look at the difference you made between the two and how they compare in terms of detail. I would encourage anyone not to stick to the same style but to keep trying different methods of drawing things, even if the first result doesn't look as good as what you previously did, it might just be because you lack the experience in that way of drawing things. Once you've tried it all, you can then choose a style you like based on what you like and how well you can draw with it. However, this does NOT mean that you should expect someone to rate something you've done later better just because you were experimenting with new styles. If a tileset isn't as good as somehting you did earlier, then don't expect a better rating for it.

Don't ever...ever...EVER rate an example level for a tileset...EVER!
If you do I will...do things that I am unable to say on this board. If deleting ratings on a tileset based on the example level of a tileset is not a rule on J2O already it sure as hell should be, because it is the one thing that tileset makers out there hate and hate and despise and hate more than anything.

Picture this: You spend a month on a tileset, you work your sweet hiney off on adding the detail, making the mask fit perfectly, making sure all the tiles are there, betatesting it with a group of people to amke sure nothing is missed out. After this time you finally make an example level just to show how the tiles fit together with one another and how they are used. After all this work you get a review where someone says: "The level needed a powerup. -1 points".

I would burst into an uncontrollable rage. While the above example is a worst case scenario, there are other cases that are less noticable where people just rate the tileset based on the example level alone. They don't look at the mask and they don't bother to look if there are any extra tiles or ways of drawing it based on the tileset image and just assume that the example level shows how all levels should be made with the tileset. This is BAD! DON'T EVER DO THIS! When you review, please pretty please only use the example level as a guide. Look at the tileset, look at the mask, look at how the parts fit together, look at how the pieces fit together, stare at that little seemingly insignificant little rock thingy and look at how much time and effort the creator of the tileset put into making that small piece of detail look as good as it is and how he made it fit into the rest of the tileset in such a way that it is so perfect you don't even notice it.
This is what makes a good tileset, not the example level.

"Don't critic sets from the way they're drawn, but its mistakes."
This is wrong, with that sentence you are saying that you should completely ignore the amount of detail in the tileset and only rate it's usability. It takes effort to put detail into a tileset and make it look good, a lot of effort, don't ignore it. However, you shouldn't just rate a tileset based on it's detail alone, do look at the mistakes, do look at the usability, but don't forget to look at the detail aswell.