Raik Postmoterm

Very interesting! I guess you’ve read my (rather hastily written) thoughts already, and there’s nothing new there anyway.

On the language question: I often think one of the main accessibility issues of IF which graphical games don’t suffer from quite so much is that it’s predominantly in English, and there’s no easy solution to this problem. Twine is pretty language-neutral though, while other IF systems take a lot of effort to port to other languages. Still a Scots game is only as accessible as it is because Scots happens to be intelligible for most speakers of English. A Welsh game would not have been able to “reach” anybody outside of the Welsh-speaking community, no matter the universality of its themes.

You are first to post a public postmortem, and a good one!

I’m very surprised my effort placed only two below yours. I liked the original, but I’d be glad to play an updated version.

Also, thanks for all the reviews in the author sub-forum. You helped make it fun and interesting.

ETA: HanonO wrote a postmortem in the authors’ forum I’m embarrassed I forgot about. I hope he shares it publicly. I think people will have a lot to say.

I did read it, Kivie, and thanks for it! It was one of the reviews that made me properly twig that I hadn’t designed it properly for all possible types of play.

Thanks for the kind words, Andrew. I had the postmortem all queued up and ready to go, oh dear [emote]:-D[/emote]

An audio version would be a great idea. Scots may be intelligible, but it doesn’t mean it’s easy, which creates a bit of a barrier for a lot of people to what’s arguably your main content (or at least, the English-language thread can’t meaningfully stand on its own in this kind of work). I said this in my review also, but for me in particular, hearing it would be a lot easier to understand because I’ve spent enough time with Scottish people to be familiar with what they sound like, and when I was playing it during the comp, I wished there was a way to have audio of the text as it was a chore to wade through (so honestly I skimmed it rather than getting the full impact). I don’t know if Google Reader would be able to produce anything meaningful, but machine-generated audio still isn’t particularly good yet, so I didn’t try it.

Thanks for this post-mortem. It is great to see the thinking processes authors bring to their games. What’s your timeline for a post-comp release?

This is a long-term pattern, preceding the resurgence of choice-based games (and Twine in particular) by some time.

  • People who enjoy writing reviews are generally going to be people who enjoy writing. Obviously writers are going to be more likely to prefer writerly pieces.
  • Critics like things that give them lots to think and talk about. This makes them more interested in experimental and unusual pieces, or pieces that intentionally evoke unpleasant experiences. If you mostly play games for straightforward enjoyment, you’re less likely to write reviews, partly because there’s not much more to say than ‘I had fun’.

Right, it’s the same phenomenon you see when you compare the box office numbers with film critic reviews, etc. I expect the Comp. ratings correlate very highly with “how much fun did you have spending two hours playing this game?”, with a slight bonus for games that experiment in a space slightly outside of what has been done before, and a penalty for games that are too experimental.

In any case I was glad to see that the score distributions for most games look natural, with very little evidence of “strategic 1-voting” that had been threatened at the start of the comp.

All the same, I would like to encourage people who think this, to actually write reviews, even if they might not be critical masterpieces. Because this can often turn into “Why didn’t game X do Y better?” or “game X should’ve been more/less like Y.” And this isn’t literary, but it is useful, whether game X’s writer has played Y. Even knowing what was fun, where, has helped me, because I say, how can I change this area? I thought I did the best I could, but maybe I can have a think.

This has worked with testers too often to be coincidence.

There’s no way to make people do this, but I think there’s a lot of worth in it, as an author, and I’d look to encourage people even if they just write laundry lists. They can be less mentally taxing than perusing the (themselves useful) online transcript. A big-picture view isn’t all bad.

Thanks for the detailed postmortem! This one for me was a case where the gap between what the author intended, what the author seems to have produced, and (this) interactor’s (sometimes rigid) demands on stuff. (As a reviewer I tend to focus very heavily on how interaction design worked for me, sometimes to the complete exclusion of all else. This was thus a very detailed test case.)

But first, trivia:

In digging around on online resources so as to interact with it I was actually struck by the similarities between various Americanisms and Scots. I assume that has to do with immigration patterns from centuries past but growing up in the Southwestern US my main reaction was a sort of mind-boggling half-familiarity.

That might be somewhat inevitable. There was a lot of stuff which had or easily could have had content warnings this competition. A prospective judge that’s actually bothered by that might start out by simply not rating those entries, but if that sort of material gets too prevalent doing so would literally disqualify them as judges. Is downranking stuff because you found the content unpleasant or inappropriate “strategic” voting? That’s a schism all by itself.

Or wondering if making comments that are innocuous when thousands of miles away end up invoking aspects of centuries-old feuds that make it look like they’re taking a side they know nothing about. [emote];-)[/emote]

The text for switching makes it sound like they’re actually the same text, as if this were a traditional dual-language book. (Entries like this have actually been submitted before.) A forced shift near the front would make it clearer that things are not as they seem, but I’d worry a bit about someone to whom Scots (or English!) was sufficiently impenentrable that they couldn’t immediately see these texts were not directly related.

Most of my thoughts on this involve the interaction design and you hit a lot of the most basic points here already…

Here’s where I get slightly dogmatic: if there is no connection between choice and result, it’s not really a choice and the player is actually faced with three next buttons and an RNG. You can’t master a pair of dice.

So if you were going for a completely powerless interactor, you failed, but that failure was a good thing, because if you had succeeded, it would have been an automatic 1 for me on the “is deliberately wasting the judges’ time to no good end” criterion. You’ve been writing about this as if you had literally code into this the Twine equivalent of the Inform 7 code “Every turn when a random chance of 1 in 10 succeeds, end the story saying ‘Hope you saved recently. Oh, wait, I totally disabled save. Ha ha!’” When I was experimenting with this, I didn’t encounter it.

What I did encounter was a maze that turns out to be self-solving, and a set of opaque options for which some had static results. This did in fact let you master it almost immediately. Here is the “solution”:* Play through, marking through which options caused you to lose HP/SAN.

  • If you end up having to replay, never take those options when it’s your choice.
    That looks like it converges rapidly to a playthrough that also wins.

Not a lot of people called out the self-solving maze as a thing because you only notice if you read the design notes. (If you happen to be in Scots at the time it also looks like it might be some kind of thing that is maybe based on your SAN score.) Having it self-solve after a dozen-odd moves is a reasonable way of defanging it. (It is, however, including a maze in a competition game, which is a dangerous move if you want to avoid antagonizing them…)

Here’s where I get real dogmatic.

But first, I want to note that in my playthrough, you basically didn’t hit your design goal: I cleared the game with one HP/SAN left on my first go. As far as I’m concerned based on a single playthrough, this game is therefore Perfectly Balanced ™. It’s only once I start treating it like an object to do science to - or reading the design notes - that I start getting unhappy.

Also, by your own statements here, this totally can’t work. Random insta-death with no UNDO doesn’t invest our choices with meaning - our choices are (as you stated above) irrelevant to actual results, by design. And it’s not just no UNDO, it’s no SAVE. So this is “occasionally you will have to restart the entire thing without warning.” You might as well randomly crash the browser at that point. That is the opposite of a meaningful choice.

Anyway. UNDO. Metaverbs generally, really, if we’re talking parser IF, but there are parallels elsewhere. Messing with the metaverbs to in some sense raise the stakes is a very old idea, going back to Adventure itself (whose original last lousy point was to win without ever saving). So this is a discussion that has been going on in some form for over 30 years at this point. I’m stating this pretty categorically but remember that I’m in Dogmatic mode here and speaking for myself…

  • If you haven’t disabled SAVE, don’t disable UNDO. UNDO can be tediously simulated with SAVE so if you’re doing anything at all where UNDO would help, disabling it is merely increasing the meta-level tedium of the act of interacting with the work.
  • You almost certainly won’t be able to get away with disabling SAVE. (The only entry this year that can legitimately do so is Begscape, due it having an interaction structure more like Asteroids or similar endless high-score-based games.)
  • It is never OK to mess with RESTORE, RESTART, or QUIT.
  • You do not control the interactor’s environment. If for whatever reason SAVE is disabled, even selectively, technology permitting there should be some way to freeze play on one machine and resume it after a reboot (or ideally even on another machine). The interactor might be on the bus or something.

It’s worth noting that Depression Quest quietly meets all these criteria, and does it so smoothly that it’s honestly hard to imitate and sounds silly to point it out: it does it by making Back, Forward, Bookmarking, History, and URL cut-and-paste all work - the normal range of browser interactions. The interactor probably won’t even have noticed that they did that, when they wanted to go back to check what a thing said, and hits back, then forward, and it works. IF players would not normally think to have a REDO command (and, indeed, technical details of Z, TADS, Glulx, etc make it not entirely meaningful), but here it is, doing its thing.

Granted, this is hard to square with something where the results are in part the effects of a random number generator. “So don’t use a random number generator for anything”, says the parser IF consensus, but even in dogmatic mode that’s a bridge too far for me. Randomness is a problem for other reasons, though.

After the interactor has done a full vertical (in the sense you used) traversal, they may want to see more. When that happens, they shift into a mode where they aim for a more “horizontal” traversal. Horizontal traversals are deadly tedious without access to UNDO a similar technology. When playing in horizontal mode, the interactor is basically running experiments on the game engine. I’m of the firm opinion that making this harder is a really bad idea, because you’re both interfering with the way people who are heavily engaged want to work with it to get more out of it afterwards, and because it’s ultimately futile. (In the extreme case, they could run the Twine in a custom Javascript interpreter that lets them dictate random number results.)

So if you do bulk this out, it gets really important to make it possible for the player to say “so what happens if I do…”? and make it doable.

As you might guess by now, I’m all for this, but how is this not just a more convenient implementation of UNDO in the absence of transcripting? (See Milk Party Palace’s wildly misnamed RESTART option for what seems an awful lot like an implementation of this.) For that matter, is there any reason to disable save/bookmarking capability?

Interestingly, I am the only game with an extremely obvious 1-spike (in a voting pattern across the comp of pronounced normal distribution), although both Krypteia and With Those We Love Alive have something that looks similar, if not as pronounced. My working theory is that this is 1-voting the Scots as impenetrable, but it could be something else. On the one hand, it’s only 5 voters; on the other hand, it’s as much as 8% of my votes. If anyone’s here who 1-voted me, I’d be really interested to hear why. I give an honest promise not to be defensive about it: I’m just interested in understanding IF’s different audiences. Feel free to PM me.

On that: I’m also not saying any of this out of sour grapes. It is, for me, about understanding the different audiences that IF is playing to as a means of making creative decisions about those audiences.

Also on this, I’d discourage anyone from complaining about the voting pattern on any of these grounds. I think it’s good to view competitions as a means to direct publicity and grow audiences, rather than a means of finding “the best”, which is a massively suspect category anyway. Which is why I’m really keen to keep comparing apples to oranges and seeing a diversity of game-forms in the comp, however the voting goes. And if the diversity of the comp grows, I imagine the diversity of its voting audience will grow too.

Michael, thanks so much for the lengthy and engaged response. I think we probably disagree on some fundamentals, but I’m really glad to have the conversation!

Yes, you’re absolutely right, and it’s really glad to hear that picked up on. I recently did a wee poetry tour of NE America, taking in Toronto, New York, Montreal and surrounds, and had a lot of great conversations about people half-recognising bits of Scots. Nova Scotia in particular has its own (struggling, but alive) variants of Scots and Scots Gaelic; my own home, Orkney, contributed a huge proportion of early Hudsons Bay workers and those echoes are still heard around there too.

Oh, inevitably. Mostly I think that Scots should get over themselves about this. Scotland was colonised, it’s true, but mostly Scotland came out of the first phase of settler-colonialism totally on top as an imperial nation, and needs to start seeing itself as such. So you are likely to encounter chips on shoulders, but I recommend offering thoughts anyway.

Hmm, yeah, it’s actually more complex than this. I didn’t want RNG effects (and didn’t include them), but I also didn’t want mastery effects: I wanted choices to feel dangerous, and half-controllable, but not always in the player’s control. If you’re careful and pay attention, it’s not too hard to avoid much damage/panic, but you will still get hit sometimes despite your best efforts. That’s how I want to model panic as hitpoints. I’m not sure how well it succeeds at this, or even if success is possible. But I’m going to try.

So this is interesting behaviour, and I’d be interested if anyone played like this. You’re right that that is totally possible, and would “win” the game. You’re almost tempting me to actually include RNG at one point to stop this happening – but no. What that behavious would do, though, is to create a major gulf between player and player-character, by making the player have material access to information the PC doesn’t have, which is an immersion breaker. I don’t have any answers here, but I’m not sure I have a problem to solve either.

Fortunately, as might be apparent, I am totally good with antagonising players. I think there is interesting material to explore in artist-audience antagonism. It’s explored in every other artform, but is coming to games a bit late. It’s increasingly common, though, which is making me more suspicious of leaning on it [emote]:-D[/emote]

Actually, I did design it with that balance in mind: a non-averagely careful and attentive player is likely to achieve this, but a statistically average player will be made to reset once, and some players will need to reset more than once.

I sort of agree, and sort of don’t. I think I’ve misled you by saying things like “arbitrary”. What I actually want is choice that is neither opaque nor transparent, neither totally arbitrary nor totally predictable. I want choice that you feel almost able to control, but never quite. That’s how choice feels in a state of extreme anxiety, and is what I’m trying to model. I agree that choice in an RNG removes all meaning, but I think if I can do my job right we end up with very very risky choice, and I do think that is invested in meaning.

My comparator here is Roguelikes. I love the feelings associated with permadeath: a sense of fatalism, and the knowledge that every decision and movement is fraught with very real danger. Roguelikes are much closer to a state of constant anxiety than, say, Bioshock, which mostly empties death of meaning.

Just for the record, this all comes as standard in Twine.

This is a really good point. The “horizontal content” that I want (is there any other language for this, or is this actually quite a good phrase to use?) is all world-building and detail-adding rather than extra danger that will require undo.

Something else that I’m adding in a rebuild, which I forgot to say earlier, is to record text progress within any given node, so that the player never ever has to repeat the action of clicking text to expand a passage if they switch languages and return – and this will matter for expanding horizontal content.

It’s a slightly less convenient version of Undo. It’s because I want to still imply the effect of permadeath, and to explore the space of repetition: the cycle of anxiety - panic attack - anxiety repeating day by day, and the horrible aesthetic of endlessly-grinding cavecrawls. I think it’s a compromise that’s worthwhile to make – it lowers the frustration barrier without removing all the experiences I want the player to have. In the end, though, I’m always going to antagonise some players and interest others, and I’m at peace with that.

I hope I haven’t come across as defensive here – I don’t feel attacked, and I’m not trying to push back! For me this is about hashing out the aesthetic effects of different design choices, and exploring creative differences and why they might be there. Thanks again for all the thoughts.

Yeah, and I think as written that’s fine. It’s worth noting that I grew up playing the old action-style arcade games as well as IF, and so some of that leaks over. The criterion I was judging by is that while unavoidable damage is completely OK, you should never actually be in a posiition where the game was effectively unwinnable from the start. (The traverse-a-map descendants of Space Invaders are judged in part by a criterion that goes something like “a player with perfect knowledge and infinitely fast reflexes should be able to either beat the game on one life, or something about your game is irredeemably broken.”) There are examples in IF too: there was an infamous printed-book CYOA called something like Horror House where you had a Fear meter and various scary things could unpredictably drop it, and if that meter hit 0 you lost immediately.

Which is all well and good, and pretty close to what you seem to be aiming for… but the reason it is infamous is because it turns out that there is a minimum amount of Sanity you would need to win under any path with presciently optimum choices… and you could very easily start the game with a fear capacity lower than that rendering the story unwinnable by any means. Its reputation did not turn out to be “oh wow, that is hardcore,” and instead was more like “oh yeah, it’s that gamebook you’re supposed to cheat at”.

This quality would be shared with roguelikes - Nethack randomizes what scrolls and potions are which across each game, but more experienced players know how to engineer situations that will regain them this information.

I personally would not be generally concerned with immersion in this case. A player playing in the manner I described is doing everything they can to break the system, and has consciously chosen a non-immersive playstyle. And come to think of it, while I’ve described the behavior bloodlessly, it doesn’t seem that improbable: we’re basically describing a player who was punished for taking actions less than five minutes ago. It is completely reasonable that the player on replay would shy away from those actions that were just punished, particularly if they haven’t yet twigged to the notion that there is a random element.

One of our fundamental differences is that I consider the designer-player relationship to be less artist-audience and more playwright-actor. This is definitely not universal, and it may even be part of my arcade background. Playwrights that antagonize their actors are not really advancing the world of theater, but sometimes the difference between antagonizing the actors and challenging them to seemingly-impossible heights is a subtle one.

That said, messing with the player in various ways has a long and honorable tradition - TVTropes has very long entries with names like Interface Screw, Unwinnable By Design, and Fission Mailed for messing with the player at the actual gameplay level, and the tradition of mocking players for foolish behavior in IF goes all the way back to Adventure (which also undercuts it - the dragon is probably an entire paper all on its own). I would suggest that if games are late to the party, it’s because they’re a younger medium in the first place.

This implies that the incredible dearth of savestate capability this comp was a deliberate and widespread choice, which is kind of alarming from where I stand.

I’ve been thinking about this more, lately, and I think it will do for now but it doesn’t hit the extremes. At the far extreme of vertical play, you would have someone who plays the game exactly once, until something causes the game to end, and then walks away never to return. This makes Super Mario Brothers the story of a bold hero who walked a hundred feet and then fell in a pit and died, the end, and feels like a straw man.

In the horizontal direction, I know for a fact that the straw men are real and walk around. [emote];)[/emote] The most extreme example I know involves an old strategy game called “King’s Bounty”. In this game you have a large world to explore, and you are finding clues as you go about the location of a treasure you must find to save the kingdom. This isn’t a plot-gated sort of clue, either; at any point the player can spend a chunk of their limited time to hunt for the treasure in an area, as driven by the clues. There is no requirement that you could prove you knew it was there; you just have to dig in the right spot.

So there is a playthrough devised of it that sits around at the menu screen for awhile, fiddling around, and then starts a game and wins in one-fifteenth of a second, because their fiddling around on the menu screen was precisely calibrated to prime the game’s random number generator (which was seeded by things like how many 60ths of a second the player spent on any given menu item) to generate the treasure one step away from the starting location, and then spent the requisite four input frames needed to take the step, dig in the location, and confirm they intended to do so.

This kind of thing is common enough that there is a name for it: “luck manipulation tool-assisted speedrun”. I have occasionally called the mode of interaction for such things “vivisection”, and it seems like it is the extreme of horizontal play.

This is, I think, the practical extreme of “horizontal play”. There’s a bright line between playing with the system and playing within it, and Raik certainly has room for horizontal expansion while staying within the system. (Something like “exploratory” might be good, but as other threads show, that’s kind of overloaded too. Emily Short recommended a book on game terminology on her blog a few weeks back and I’ve been meaning to check that out.)

This is a really good idea and not just from a gameplay perspective. This sort of behavior is something I’d expect from any interactive program. If I said “I want to look at something else” and then “take me back where I was”, it better be back where I was!

Mmm. I honestly didn’t consider that angle. So it might be personal bias but I think it would be best to make the in-world repetition more explicit if so. If I hit RESTART, my thought is not “next verse, same as the first,” it’s “right, that didn’t happen.” If I’m deploying a “meta” verb, I’m out of character and am commanding the system somehow. (This is why disabling RESTORE or QUIT in a parser game is so unforgivable; it’s asserting sovereignty over the player’s machine. Disabling SAVE is an out-of-character annoyance, but disabling RESTORE is an out-of-character threat. Or would be if intentional. All cases I know of that did this were clearly bugs.)

Making restart-and-work-through-the-day-to-the-last-point work cleanly seems easy in the Scots world, but harder in the English one. One possibly amusing thought there would be to have slightly different hero in each playthrough in the heroic fantasy side - a freshly randomly generated roguelike character.

A sticky question that comes up too: what happens when you replay maybe-bad commands? Do they stay the way they were the previous day? If not, do we risk essentially dying during the recap? That would be clumsy. Rewinding to just before the last always-bad action seems like it might give away too much of the mechanics. But if so, we seem to be losing some of the unpredictability you aim for.

And yeah, if it isn’t clear - I’m not trying to “attack” here in the usual sense. I’m also interested in the aesthetic response space, albeit from a different standpoint. But that does come out a bit like a fencing match. [emote]:)[/emote]