Stuff not to do as an author, a short list:
These aren't all the possible errors a beginner could make, but these four were the ones brought to mind after playing Deephome. Deephome: A Telleen Adventure, by Joshua Wise, is more or less a standard fantasy game. Oddly, the flaws in the game are not related to genre — the story isn't completely well thought-out but it is fairly original. Instead, the problems I had with the game were classic design flaws such as the ones mentioned above.
The first was, I think, the most serious. There were about sixty rooms in the game; of these, about a third had no purpose whatsoever. If you were to take out the extra hallways that lead to the useless rooms, you could probably reduce the number of rooms in the game to about thirty without hurting gameplay or story at all. The author apparently attempted to tack on a rationale for visiting all the rooms by telling your character that he must check out all the rooms to make sure they're safe. Which is great, but doesn't make it less annoying.
There are a couple reasons not to have a big empty map like this. The most obvious is that the player has to walk farther. But another, somewhat more subtle problem, is the more roughly-sketched areas you implement (and most of these rooms are fairly broadly implemented), the more obviously this simulation of reality fails to match the real thing. For example, at one point in the game you need a weapon and armor to kill a monster. There is an armory in the game which has both, but you are not allowed to take any. Instead, you have to find the armor hidden elsewhere. Why did the author do things this way? Or another example — your player needs to forge something. There are, let's see, four smithing rooms in the game. Only one of them can actually be forged in. Why why why? And on a somewhat related note, the more rooms in the game, the harder it is to find the important ones (not to mention that encountering a series of useless rooms encourages players to ignore other ones and hence miss important details).
This of course ties into the "don't make the player read your mind" rule. For example, at one point in the game there is a monster who is allergic to some plants. There are lots of possible places plants could be in the game. Only one of them has a takable plant that works on the monster. As far as I can tell there are no clues whatsoever to figure out which plant to use, or even that the plant in question is takable. In a smaller game you could at least just try a brute-force solution but here even that is impossible. This same situation is repeated elsewhere in the game when looking for a fire source. The size of the game magnifies an existing game-design flaw.
The size of the game also makes it annoying to walk back and forth. There are two reference books in the game. One is takable and the other one isn't. At various points it is necessary to look up some information on the various creatures in the game. For some of them, you must 1) find out what to look up 2) walk all the way over to the non-takable encyclopedia and look it up 3) use the other encyclopedia to look up stuff from that lookup 4) go back to the creature and use the information. Why design like this other than to piss off the player?
Finally, the verb thing. This game is by no means unique in not replacing library responses, but I wish more would. Especially with Inform games, where the library's text is basically oriented towards writing Curses and nothing else. Still, "Violence isn't the answer to this one." is frustrating when you're supposed to be a hardened dwarf warrior, especially when it's in a situation where the action is completely plausible. That by itself is annoying, but it turns from annoying to hateful when this is a guess-the-verb situation, and although "hit" doesn't work, "beat" does, so the default library response to "hit" pushes players in completely the wrong direction to solve the puzzle. One final gripe is about the parser. Authors have to be very careful about snarky comments from the parser. Often these aren't funny anyway, and even when they are, it only annoys the player if the real problem is parser stupidity rather than player error. Usually better to be safe than sorry about this, I think, especially if the game is more or less serious.
Of course, Deephome: A Telleen Adventure does have some redeeming features that I hope will be carried into the sequel. Like I said, I thought the fantasy setting was decent. Ok, there were dwarves, but at least there were no dragons. In fact all the monsters were reasonably original. Wise seemed to have the right ideas about what would be plausibly found in this sort of setting, even if he felt he had to put everything on the list into the game. There's a trick to good writing that involves suggesting a large world but only showing a small part of it — this often carries over into IF.
The magic was really pretty good; the ingredients were a little goofy but the design of the casting was nice, and I thought the writing really captured what the author was trying for. It had a maze; that sucked. But the other puzzles all felt integrated into the setting, not thrown in for kicks. Oh, and there was a rail car. I love little track-running vehicles in IF games. They're fun to ride around in and they do a good job making the areas feel bigger. Plus this one was implemented especially well, making the controls easy to read each time to enter.
In summary, were Wise to write this game over, he should cut out half the rooms and bulk up what's left. As it is, it's too thinly spread to be fun to play. The webpage suggests a sequel in the same world; it'll be interesting to see if the gameplay improves to match the settings's potential.