[initially appearing on rec.games.int-fiction, on the 8th of December 2004]


Hello all.

Well, the dust has settled on the 2004 interactive fiction competition and the stream of reviews has dried up. It thus seems like a good time to look back on All Things Devours and answer some of the questions and comments raised in the reviews. Since there were so many and they dwelt on several common themes, I thought I'd do it all in one place.

I guess that a logical place to start is the origins of the game. There was quite some speculation on this in the reviews:

Sam Kabo Ashwell wrote:

> First thoughts: my, somebody really liked Spider and Web,
> but instead of coming up with some sweepingly original
> narrative techniques and deeply clever puzzles,
> they just nicked the basic story

Andrew Krywaniuk wrote:

> (it certainly owes a nod to Spider and Web or Moments
> out of Time, as well as several popular movies)

Timofei Shatrov wrote:

> There is some interesting stuff going in this game which looks like a
> Vicious Cycles clone.

Actually, it is a bit embarrassing to admit, but the only works of IF I had completed before the competition were:

Shade
Galatea
I-0
9:05
Common Ground
The Frenetic Five vs. Mr Redundancy Man


And I had investigated but not solved:

Zork
Hitch-hiker's Guide to the Galaxy
A Change in the Weather
Lock and Key


So no, I have never seen Spider and Web, Moments out of Time or Vicious Cycles. I don't really know why people were so quick to claim that I had. As it turns out though, two of the works I had played did directly influence me. A Change in the Weather's 'about' text made it clear that try-by-dying was still acceptable if done right and that encouraged me to proceed (as well as to warn the players via the 'about' text). Common Ground was a more direct influence, as it was there that I saw an attempt to record past player actions and use them again. In particular, there was a review of it by Ian Merrick which featured the following:

> I don't think something along the lines of Back to the Future
> would be possible in IF, though, where certain scenes are
> viewed several times without distortion, and where the plot
> depends on every tiny event meshing together perfectly.
> How on earth could you do all that and give the player
> reasonable freedom of action?

Well, that seemed to be a fair challenge, so I thought about that question for a while and thought about how cool such a thing would be. For what its worth, Back to the Future and its sequels did leave their mark on me and they would certainly be an influence. I also have a fair interest in science-fiction and had thought about time travel quite a lot in the past. (While on the topic, I would recommend 'Thrice upon a time' as a classic of the time-travel genre). I then thought of some puzzles which became the batteries/door/alarm puzzles. That night I downloaded the Inform libraries/compiler, the IBG, and the DM4. I Hacked the first IBG toy adventure to record the player's movements, saw that I could do it, decided I'd write the game and went to sleep.

Of course there were many elements of it that were difficult to code, many of which were probably a bit unnecessary. For those who are interested, I will release the source code with the updated version, so you can see how it was done. I don't promise that it will be too comprehensible though, so sorry about that. All in all it ended up taking two and a half months, on and off, to write it, which was longer than I had hoped, but not altogether surprising.

I taught myself Inform by getting the automatic and manual doors to work smoothly, since the ridiculous default nature of doors in IF is really annoying. Everyone should at least copy and paste in the door class from the DM4. Ideally, Inform's library should be changed to do this by default. The current way is just terrible and does a good job of both spoiling any mood created as well as turning off new players. I am glad that I got a few positive comments about my doors, but it really should be the norm. That said, it was embarrassing that I somehow introduced a bug on the day before the competition and hadn't retested the useless doors. Interestingly the bug only occurs if you try to enter them, something Inform generally does not allow anyway if they are closed. It was only my tricky door programming that led to this bug and only my tricky door programming that made people find it (since I make entering a door without opening it work elsewhere).

I enjoyed writing the text, although I did so much faster than I would have liked as the coding took so long. I may write a little bit more for the winning ending, since many found that anticlimactic, but I am not sure. Ultimately it is a game that doesn't have a natural endgame and I think this hurts it a touch, but don't know if anything can be done about this.

Quite a few people said that they liked the terse and tense phrasing and atmosphere and the general conclusion seems to be that it was sufficient or slightly more than sufficient for a puzzle based work. On a somewhat harsher note, Jake Wildstrom wrote:

> Highly flowery writing; it's not adjective-poisoning, but a subtler
> form of overwriting. Perhaps it's the personification of everything
> that does it.

This is a fair comment and I would probably agree. I just didn't have the time to strip it back a bit more.

The response to 'look at myself' (or, as it has become known 'x me') was surprisingly strong and diverse. For those that don't know, it reads:

> Dark hair frames a warm face. Despite the anxiety written now
> in your tense brow, the confidence of youth and vibrancy
> of intellect are clear. There is a depth in your clear green eyes.

I knew this was often the first command, so I had spent a little while writing it. On the one hand, some people liked the portrait I gave, but there were also a few other reactions. For instance, Mike Russo wrote:

> The PC description admittedly
> made me roll my eyes a bit, both because it's slightly extravagant and
> because the gorgeous female MIT grad student seems like a bit of a
> wish-fulfillment character, although of course such do exist Å0É2 I
> certainly met more than a few during my time at Caltech. And
> portraying her as dowdy and asexual would have annoyed me as well, for
> buying in too strongly to the stereotype. Perhaps I just find the
> depiction of female sciencey characters somewhat problematic in
> general; it's hard to sail the ship of characterization between the
> poles of stereotype and pandering.

Now I understand Mike's point and similar issues occur with many depictions of stereotyped groups. Often people do overcompensate. It is common to see an attribute attributed in a stereotypical manner and then want to reverse it for your work to overcome this, but this can lead to silly polarizing of the issue. In a way I was conscious of this while writing All Things Devours and decided that the protagonist would be female, but that this would not be remarked upon within the game. I am a grad student myself and almost all of my friends are too. Many of these are women. This perhaps gives me the benefit of being rather unfazed by the stereotypes and just writing the story. However, when it comes to the 'gorgeous', 'wish-fulfillment' character, I'm afraid I have to disagree with Mike. If your read the description I gave again, it says nothing about Natalie (the PC) being gorgeous or even about being conventionally attractive. She has signs of warmth, confidence and depth. While these are things I find attractive, I am hardly going down the 'supermodel as scientist' route.

Others commented on the strange perspective of the description. For instance, Zarf wrote:

> Only a nitpicker would point out that the self-description reads like
> someone *else's* description of the protagonist -- not like something she
> would say herself.

This is a good point. Now when I type 'x self' I don't imagine the character actually looking at himself/herself, but rather see it as a kind of meta-command. I think of it more as a third personal perspective, letting the player know a bit more about the character that they are trying to roleplay. In this, I perhaps find things to be a little like those third-person 3D games where the perspective is right behind the character. You act for the character, but see them externally. Or perhaps it is like the descriptions on a character-sheet in roleplaying games, which are generally written from the third person. In any event, I asked Zarf about this and he agreed that it was a kind of meta-command, but for him it is really asking for the character's thoughts about their own appearance (and presumably not to actually be sitting there thinking about herself for a turn, either). This probably fits more with the (few) responses I have read for this command and I am prepared to agree here. Still, I'll leave this as it is for the next release.

There were also many comments about the title 'All Things Devours'. I had wanted a title that gave an impression of darkness and the dark side of the passage of time. I had also decided to avoid all explicit references to 'time' and 'time-travel' in the game, and thus was looking for a way to reference 'time' obliquely. I remembered this poem/riddle and found 'all things devours' to be the most evocative phrase in it. I was very surprised to see that several reviewers thought it was a typo and this says a lot about the expected standard of competition entries. I was also surprised that many people commented on the fact that it was by Tolkien and on how this was a problem. Myself, I don't see the relevance of this. In any event, I stand by the title and like it a more than those of most of the other competition entries. Each to his own I guess.

I guess I should also mention that when the poem appears upon your failure, it was originally interleaved with the prose in a manner that I found quite evocative and perhaps reminiscent of Simon and Garfunkel's Silent Night. I liked the idea of trying to achieve this effect in text and to the best of my knowledge no-one else had done so. Then my beta tester informed me that it appeared prominently in last year's winner, Slouching Towards Bedlam. I hadn't played Slouching, so I tried it for a few minutes and found it excellent (much better than Devours, from what I saw). I therefore decided to remove the interleaving, although this removed some of the power from the ending (in my view). Maybe I'll tinker with this a bit more for the next release. In any event, if you want to see it, type 'interleaving on' in the competition version, then fail in your task.


Reviewers seem to have been very sporting about the timed nature of the game, the learn-by-dying puzzles and the lack of warning for the unwinnable states. It is a credit to them that so many persevered even if this was not only an unfavoured genre, but actively relied upon three of the cardinal sins that graced many early works of IF. Obviously the timing factors were crucial to the game and it seems that most people agreed that it was worth that price. The lack of warning for unwinnable states was similar. If I could have implemented a 'winnable' verb, I would have, although it still would not have done much good: it is just the nature of the puzzle.

I theoretically could have kept the timing issues while removing the learn-by-dying aspect. I'm not sure how, but it should have been possible to go some distance in this direction. On such matters, Paul O'Brian gave a very interesting account of the 'accretive PC', a concept of a PC where the player learns by dying but the PC does not. In the model, the player just learns the expertise that the PC should already have, allowing the player to slowly learn how to fill the PC's shoes over many games. I like this idea and it was interesting to see that All Things Devours does fit this model except at one point (which is actually an optional point) where Natalie is surprised by something that, were the accretion model to fit, she must already have known. I do think this is a neat way of looking at some learn-by-dying games and am happy mine almost accidentally achieved it, but I really wasn't thinking this way on writing it. I just really liked the impossibility of the task at hand -- the way that Natalie has almost no chance of success and the real irresponsibility of her actions when seen from the player's meta-level (they are quite responsible and noble from Natalie's perspective). I thought people might enjoy the absurdity of destroying Boston over and over again too, while trying to get everything to work just right.

There were several technical problems that people encountered. I have already mentioned the door bug, which was the nastiest although not encountered by most people. There was also the absence of any 'wait until' command, which I hadn't realized would be so sorely missed. There is a prominently displayed clock in the game, so I thought hitting 'z <return>' until it got to a certain time was easy enough, taking about five seconds to accomplish, but others said it was 'highly tedious'. It will appear in the next version. On a similar level is the issue of meta-actions such as 'about' taking a turn. I will fix this too.

A more serious issue is that of poor ambiguity resolution in situations involving multiple identical objects. In the (important) situations where this is needed, it always seems to choose the worse of the two, making an additional puzzle of avoiding this. I will fix this.

There were also a few things that should have been clued better. For example, the alternate 'obvious' solution to the door puzzle where the game should tell you that it doesn't work and why. Similarly, some people did not appear to realize that the bomb is not initially set, so I'll try to make this more obvious. I will also hint at the location of your lab and a few other minor things.

As mentioned in the introductory text, I intend to release a post-comp version with slight changes to the puzzles. Some have expressed concern that making it harder would be a mistake. I had indeed planned to make it a little harder, but will probably do so with an option of some sort. I thought that some people would enjoy the puzzle-world and want to explore it for more than the 1-3 hours it seems to take most people, and while there are currently quite a few ways to complete it, making the constraints a bit tighter will hopefully make people find the other solutions (unless they initially happened upon the solutions that still work, of course). But that said, I will pay heed your warnings about how much puzzling the average person can take. I have been very busy since the start of the comp, so don't expect the post-comp release to come out too soon. My best bet is sometime in January.

Overall I am very satisfied with how everything turned out and my place in the competition. A rather hectic schedule recently has meant that I have not played the entries that beat mine, but I look forward to doing so. As you can probably see, I haven't experienced much IF at all, and have a lot to catch up on. I think I will start with the universally acclaimed ones, particularly those that will fit into small slots of free time. I think finishing Slouching would be a good start.

As you can see, I am not so qualified to comment on Jennifer Maddox's kind opinion that I should have won. Maybe. It would have been a big surprise though. I received a lot of 10's (more than anyone else?) and that makes me very happy. This was exactly the type of game that I would have *loved* to play, particularly when I was a bit younger. The 9's and 10's seem to indicate that quite a few people got that kind of intense excitement from it, so I am most satisfied with the response.

Will I write any more IF? I don't know. It was quite enjoyable to write, but just took so long. I will only write more if I stumble across another project that captivates me. Regarding the puzzle side of things, that 'ponder' idea came quite close, however, I think I would rather write something more literary. In particular, I think that one of the strongest draw cards for IF is when the player performs meaningful actions from meaningful motivations. For example: telling someone that you love them, apologising, breaking a promise, ending a relationship... these types of things that involve a strong connection between the player and the NPCs.

This is not easy to do, particularly when it comes to creating serious meaningful motivations in the player. One would also want to allow the player not to do the act and to reach some kind of resolution that way, to make it a true choice. Remember the outcry against being forced to repeal gun control laws in one of the comp games? This enacting of a law is the type of meaningful action for which reading about someone doing it is very different to doing it yourself. However, it was not well motivated, nor a true choice, and thus it led to a lot of hostility. Still, this shows how much emotional impact these things can have. Common Ground also has an important choice at the end, but with insufficient motivation making it feel a bit strange. Finally, Galatea had sufficient connection with me that I would have liked to make such choices regarding her, but the endings seemed to just happen without too much deliberation on my part. Still, to have such a connection with a character was just beautiful. If I thought I could do something like this I would be very tempted to try.

Till then,
half sick of shadows.