The last light has gone. The stars are coming out in the black sea above. Many are hidden by ice-fingered winds. My father is still not returned and the fire is almost gone.

But this is how life is: always an edge. A thin sheet on a diving-deep pool.

I hope he will return soon. I cannot summon him.

A Colder Light is now available to play online.

A Colder Light is my first released game since 2009′s Make It Good and The Shadow in the Cathedral. This one is considerably shorter and easier than both of those. It’s also my first text adventure to use no keyboard input. It’s a short tale of magic, courage, animism and ice.

No save is implemented, although the game is short enough that you shouldn’t need it.

The game doesn’t work nicely on phones, but should play okay and runs dog-slow on an iPad.

Comments in the comments, and bugs to the address in the help text, if you please!

So what? you might think. Most books are read once, most DVDs are watched once, most Christmas cakes eaten once… But I don’t mean finished, I mean played. The majority of console games are opened, installed, booted up, played for a single session (possibly of several hours), then never booted up again. Even though games can afford tens of hours of entertainment; and even though games cost four times as much as books or films.

And that isn’t true of books, or DVDs, or Christmas cake. So why the difference? Is it just because people can get stuck on games?

I don’t think so. I think it’s deeper than that. In fact, I’m not sure there is a difference between the consumption pattern for a DVD, book or a game. I think instead that the difference is in what we mean by the word finished. (And, what is inkle going to do about it?)

The length of a game can be a tricky thing to measure at the best of times. Many games provide a variety of play-modes, from different difficulty levels through to different characters, storylines and abilities to try. Some games don’t even end, like the eternally popular multiplayer side of games like Call of Duty and Battlefield.

So when has a player “finished”? I’d argue a player finishes when they’ve learnt all that they expect to learn: that is, when the game ceases to surprise. The moment when every damn level is the same as every damn level before, that’s the moment when the game is finished. From there, a player either trades it in, or keeps playing in that trance-like addictive state that parents find so worrying. (Or in some rare cases, keeps going to see what happens in the story. But those cases are really pretty rare.)

The best evidence I have for this idea is from playing Assassin’s Creed II. The player takes the role of a member of a secret assassin’s guild in a rich simulation of medieval Florence and Venice. The cities are modelled on their real-life counterparts, and the player is given the task of navigating their streets, crowds, rooftops and canals to complete a variety of missions against the powerful ruling elite.

It’s a game which is remarkable for being almost entirely one long tutorial. The game features a staggeringly large number of things the player can do, but they’re all introduced gracefully so that the game is still teaching new mechanics at 70% complete (the pistols, the extended jump). By that point  most normal games have settled down into a familiar rut and are simply ramping the difficulty curve.

This constant learning made for good fun – each new mechanic introduced a new challenge, opened up a new route, or provided a new way of interacting with the simulated crowds. But despite the breadth, most people I know gave up on the game without reaching its end. And most gave up around the same point – the Venetian street carnival, an extended sequence of (none-too-difficult) challenges in which no new mechanics were introduced. People said, “They felt they had seen all the game had to offer.”

Assassin’s Creed, a very beautiful game with remarkably few assassinations in it, is a great demonstration of the parallel draws of novelty and mechanics. The mechanics in the game are not great: the stealth is fussy, and the running away sequences poorly balanced and quite repetitive. The first game in the series relied on these mechanics, and it got quickly very dry. But the scope of the world in the sequel, and the amount of novelty it is able to offer, is very large, because of the breadth and depth of its simulated cities and citizens.

By way of a quick contrast, Valve’s game Portal introduces a novel concept (creating spatial portals linking the world together in Moebius-strip fashion) which has lots of subtle depth (arising from momentum, angle, timing, that kind of thing), which it then develops with even more novelty in its level design. It’s a great example of the two aspects of a game working together in a mutually supportive way. And most people, I think, played Portal through to its end. (In fairness, of course, it was short, rewarding, and superlative.)

Most games are built squarely around repeatable mechanics – that is, a limited set of things that the player can do, which cause reasonably predictable consequences in the world of the game, with modifiers depending on the luck and skill of the player. It’s certainly true that a game with good mechanics will last, and a game with actively bad mechanics will fail.

But I think there’s another way, and that’s the direction we’re taking over at inkle for our Frankenstein project. A game – or rather, an interactive experience – can instead have simple mechanics, and be founded more on its content, and on the novelty and discovery inherent in that content. You know; the way a book is. It can be built in such a way that to be finished means to be completed rather than merely to be understood.

We hope that when people read/play (“relay”?) Frankenstein, they’ll all get to the bit where it says The End. (And then they’ll sit back, sigh, think, ‘what a great story,’ and post or tweet a nice review someplace.) If they don’t replay it that won’t be a fail. But if they play only the first chapter, think, ‘I get it,’ and put it down, then we will have missed out mark.

On a few months now until we find out!

I had the puzzle structure worked out before I coded a single word. I’m now 80% of the way through, but then I got distracted, adding hyperlinks.

I just turned off the actual text prompt thing. It seemed so… retro. There are just these buttons now. It feels kinda okay.

Building on my IF and Undum work, inkle will make interactive stories for mobile devices, and do so in a beautiful way.

Our first project, Frankenstein, is being published in association with award-winning London-based publiser Profile Books; and is being written by Dave Morris (author of Mirabilis) and Jamie Thomson, super-talented writers with a long history of gamebook work.

Right now, the visuals are under wraps, but if you want a taster, check out our interactive press release.

There’ll be a lot more features going up on the site soon, so please follow the blog if you’re interested, or watch us on Twitter.

Currently, the intelligent commands peter out after the first few turns, because I’ve not gone through the Inform 6 code adding in links to every object reference (sigh)… and it’s horrifically slow… but it’s still interesting, I think.

My previous cut of this prototype had a side-window, containing various panels – a notebook of clues, an inventory, a map (though I never got as far as building that). This version strips all of that away for a pure one-screen experience. I think, having tried both, I prefer this.

It keeps the player focussed on the one place they should go for interactions. I wouldn’t mind a pop-up hint window or something for when they get stuck – but in my more cluttered prototype, I didn’t like the way you had to actively play in several different windows to progress.

(I still want to add a graphical map. But Quixe can’t do images, yet, and I can’t draw at all well…)

Note, this is the demo with Interactive Parsing, so that’s there too, if you do decide to type. Playing the game you can flip between the two modes using that little (X) symbol that appears when you’re focussed on an object. (Erik’s had a graphic; how he did that, I have no idea.)

Lots to be fixed: I want to remove the hints in the subject line, make the suggested actions contextual, and have a think about how to do conversation. But still… but still…

(Oh, and on a Mac, the font is considerably nicer than on a Windows machine. If anyone can explain why, I’d be grateful.)

It’s a really good piece of work and shows real potential for making text games accessible: teaching the syntax while letting people get on with the game. But it also highlights one of the text games major problems – there’s way too much choice.

The strange thing, is that by including clickable links for things you can do, the problem gets worse, not better. Normally, when I play a text-game, I’m aware that at any moment I could try a huge range of different things, but I ignore most of them.

After all, most are meaningless in context. I certainly don’t work through any kind of systematic list in my head: rather I play half-strategically, and half-intuitively, and the best moments in interactive fiction feature a little of both kinds of thought.

But as soon as there’s a link for every option – even using Erik’s neat system of focusing on an item in order to offer that items functionality – it starts to get overwhelming. The contextual clues of what might be sensible get over-ruled by the visual clues supplied by the familiar-looking hyperlinks. There’s nothing on-screen to tell me that one choice is better or worse than another, so they all look on a par with each other.

And suddenly, I can’t choose.

What’s happened, I think, is that the normal text-game interface, which is analogue, has become digital. I’m no longer forming commands in a fluid, seamless way, by typing them: I’m now looking at a list. The result is something much easier to play, much but harder to read.

Looking at it this way, it suggest that normal interactive fiction has more in common with console games than you might think: both have quite wide input spaces (in console games, this freedom is in the detail of movement, camera, and aiming) but disguise them with very generic inputs – the letters of the keyboard; or the left stick and sticks on a gamepad that both have a multitude of uses, especially once combined with the shoulder buttons.

Console games are obviously analogue: text-games less so, but compared to clicking the options – or even compared to a game like Monkey Island – they have that analogue element to them.

So is Erik’s demo flawed? I don’t think so. I think all it needs it a visual pass to put the missing context back into the options.

For example, at the moment, all interactable nouns are highlighted as links. They probably don’t need to be, and it’s kind of distracting that they are. What if the links only showed when moused over? Would the player be left pixel-hunting, or would the interactions they expect to work, simply work?

What if the box of options was closer to where the input happened, and only appeared when necessary? Those of you who use Gmail will have seen the new layout of labels, which slides in and out of view when needed. Perhaps we could do something similar?

I certainly think we could use a “hidden until needed” approach for the inventory. It’s great to have it up there at the top level, removed from being a command, since the command is so utterly opaque.

It’s all about making the choices a little bit more analogue again: these choices when the cursor is over here, those choices when they’re over there. It’s about letting the player explore their interface, rather than be presented with it like a twenty-page restaurant menu.

None of this stuff is easy, of course. But it has the potential to be so very, very slick.

It’s been hard for me to pin down exactly what I want to say. The normal rules for talking about game design don’t really apply – the audience will be publishers, editors and writers, and I think a standard design talk about risk/reward and challenge/learning might send people to sleep.

Instead, it got me wondering about what gamification, and the insertion of “game-like” elements in other kinds of product and experience, really means. It seems to me to be less about making games, and more about making something that resembles a game and reminds “players” of the fun they’ve had elsewhere.

Or maybe it’s about holding people’s attention. Games are good at getting players to concentrate, and become immersed in the game-world. Players find it easy to take games “too seriously”: a game is a kind of virtual space, even if there’s no 3D-modelling attached.

That might be an attractive concept to explore, but there’s a paradox at the heart of it: publishers print books, and books are already good at getting readers to concentrate and become immersed in a virtual world. What can games offer that books already can’t?

And that the question that’s grabbed me, and where I think I’m going to root my talk. How are games and books similar, and how are they different? What, in the end, makes a game a game? Which is not same as the usual question behind a design talk, which is what makes a good game.

So I’m going to present an argument for a game needs to be a game, and hence, what makes a game-like experience feel like a game. I hope people find it useful, and after it’s done – and I’ve enough feedback to decide if it made sense after all! – I’ll write it up here.

It does it by playing through every possible game the player might have, and writing them all out separately… which turned my first example game Flaws from a 40Kb sourcefile with 40 paragraphs of so and 4 true/false flags into a 600Kb HTML.

The other day I thought; I wonder how far I can push this thing?

So I started writing a murder-mystery. With clues, and evidence, collected up in a notebook. And a branching order of investigation. It’s 163kB of source – 200 paragraphs, and 70 true/false variables. And I ran it…

Imagine this, 15 times over

…and my computer made a very funny noise indeed…

…and the battery meter went from “4:20″ to “23mins”…

…and out came a 40Mb HTML file containing 65,000 paragraphs.

The game is called A Job For Life, the interview mechanic is a streamlined, improved version of how I think LA Noire should have been, and it’s got some exploration and a even an action set-piece. It’s a purely choice-based game, with no (visible) stats or randomness or dice-rolls.

If it was published as a book, it would be the height of a house.

And I was assuming that was that, and the only way I could distribute it was to send people memory sticks like I was in MI5, when I discovered that ebook compiler software uses compression. In fact, it has a compression switch designed for dictionaries and encyclopaedias. It has a quite good compression, which got the file down to a slim-line 4Mb. Which the Kindle itself can handle, without any lag or issues. (The game’s HTML file is so big that it crashes most browsers, but not the Kindle).

In fact, the reader would never know, until they try to use the “go to” feature which tells you how many pages there are.

The game’s in testing at the moment, a process which is made worse by the way trying to compile the game takes quite a long time, as it has to play out every possible path through the text. But I hope to release it soon: with an online demo of the first chapter, and the rest available from Amazon. Wifi recommended.

For those who are interested, here’s my list, also posted on the forum.

1. Curses – While I don’t think I would ever really recommend this to anyone else, I had such an enjoyable time playing it that it has to be top of my list. I don’t think anyone has topped Graham here for his ability to turn interactivity into a conversation between player and game, with the successful solver providing the punch-line to so many jokes and having so many moments of real, intuitive insight. The puzzle of the romantic poetry book and the hedge maze are gems, that no-one would be allowed to get away with in the “real” world of games. Magnificent.

2. So Far – Plotkin has always had a wizard-like ability to turn code into world; so that even though So Far is brutally difficult and so easy to break, it never feels inert. From the (unnecessary?) pole-licking to the monster-fight in the arena, So Far felt like a living breathing world in which I was the ghost, drifting from place to place. And the ending of this was alive. Wonderful — but again, very hard, and very hard to truly recommend!

3. The Witness - Of all the Infocom games, this one was my favourite, because I actually found clues, I actually followed them up, formed hypotheses and eventually cracked the case. It took a lot of replay and a fair amount of luck, and when I played Deadline later I found it impossible, unforgiving, and over-wrought. But The Witness seemed just right to me – simple enough to be accessible, responsive enough to provide a narrative. A great game.

4. Rimworld – (I think was the name.) In the early days of the internet, a few text adventures floated around, that have been largely lost. This one was a standard collection of plastic-purple-squares and plastic-purple-slots, but back when I played it, there were no walkthroughs, no forums, and no authors emails; so I wandered, alone and without help, through an empty alien world, and every discovery was my own. Games will never feel like that again.

5. Ribbons – one of the Art Show pieces, full of connections that might or might not be meaningful. This one was great for me because it made me realise, finally, that interactivity is what happens inside the player’s head, and that what happens in the game-code to enable this interactivity is merely academic.

6. Shrapnel – bonkers, devastating, and Cadre at the height of his powers, creating a seamless experience bristling with meaning and consequence. This game for me marked the absolute heyday of the indie community; when games were quick, dirty, but wickedly effective.

7. Spider and Web – this almost doesn’t feature because, in truth, I didn’t enjoy playing one little bit. But the twist was fantastic, and the conversation system instructive, inspiring and, oh, yeah, really cool.

8. Lost Pig – Lost Pig was great.

9. LASH – I like all of Paul’s work, but this one felt the most solidly built and meaningfully executed.

10. The Weapon – great sci-fi story, with a tight design and great pacing.

11. Starcross – a masterpiece of puzzle design on a budget.

12. Christminster – this one seems to get forgotten about, but looking back I feel like Rees’ invented an entire genre of pacing here: the game is so graceful in ensuring that your scope is always small enough to be playable, but your involvement just gets deeper and deeper. I like to think of The Shadow in the Cathedral as something of a design homage.

13. Plundered Hearts – while not as tightly designed as some of its successors, Plundered Hearts managed to tell a real story and not let its puzzles get in the way, and that was a novelty in the Infocom games. And it was a great romp.

Honourable mentions: Deep Space Drifter, Leather Goddesses of Phobos, The Hitch-Hikers Guide to the Galaxy, 9:05, Dangerous Curves, Slouching Towards Bedlam.

Interactive Fiction has long battled with the problem of stuckness, and these days it’s rare to see a polished game released without hints, walkthroughs, or such an incredibly linear storyline that pretty much anything you do will work. But could we be doing better? Consoles games increasingly are trying to resolve this problem: is there anything to learn from the experiments being done in the console world?

These days, unstuckness is being heavily invested in. Partly because getting the difficulty curve right is a really difficult problem in a game of any complexity; partly because publishers want to expand the market to cover hard-core gamers and naive newbies; partly because the first response of your audience is, in the age of Twitter, the most important marketing tool you have. And finally, because the last thing a console game publisher wants is for you to take the game back to the shop in its opening release week and trade it in. They make better sales the longer the game stays in your house.

Stuckness in a console game is usually a matter of not being fast or accurate enough to do whatever fiddly button-based task the game is asking you to do. In solving the stuckness problem, most console games offer you – after a few fails – the chance to dial the difficulty down, Hard to Normal, Normal to Easy. The game will then stay on that level as you continue, and in some cases, won’t allow you to go back up again after you’ve got through your difficult patch. It’s like being put down a set in school.

The advantages of a system like this are it’s clear, it works to get you through, and it should prevent you getting stuck again, so the overall amount of stuckness experienced by consumers goes down. There’s a baiting element to it, too: by asking the player to agree to dial down their difficulty you’re asking them to admit that they’re beaten. With a bit of luck, that persuades them to keep playing just to beat the thing. When they do, there’s more achievement and as a nice side effect, they played your game for longer.

It’s a pretty good mechanic, but with two unfortunate side-effects. The first (forgive me) is the impact on the story. If the challenge is part of the narrative, then the replay-but-easier tends to negate that story beat. If Kratos sails through his battle without difficulty then it wasn’t a struggle against a colossus any longer. If Drake takes out a hundred guys without breaking a sweat he has no right to curse, “Not again” when the next hundred appear. The challenge, in an action game, is the story. Remove the challenge by dialling down the intensity of the player at the pad, and the story is no more.

Secondly, there’s a pratical problem: for the poor player who really can’t get past a certain scenario what you’re offering by replay-but-easy is a lose-lose arrangement. Either they play the section again, and are defeated, again; or they play the section again on easy, (and really hope they don’t get defeated again), and when they get through it comes with little or no achievement at all. (In fact, easy modes are often so easy that even the worst players can sleep through them.)

For these players – generally, the ones you’re trying to reach out to in providing the easy-mode-option in the first place – the experience is turned from something thrilling into a chore: a lesser experience with a lesser reward. And we know that’s true, because there’s something in the game design mentality that means when the idea is suggested people think, “but fair enough: after all, they couldn’t do it.”

We should remember: as designers, we are entertainers, and our purpose is to entertain. If our game is too damn hard, that’s our fault. (In software, the old adage about bad workmen and tools has never been true; in games it’s similar: a bad designer blames the player.)

LA Noire (did I mention I’ve been playing LA Noire? Thought so) takes an interesting different path. After failing an action sequence a few times – a car chase or a shoot-out or a fist-fight, though failing one of those would be pretty embarrassing considering how poor the fighting mechanic is – the player is offered the option to skip the sequence entirely. The screen even goes to the lengths of pointing out that skipping will not affect your score or your progress. It’ll have no consequences except you get to move forward with the game.

Classic game design theory states that players will always click yes. Players game and cheat the system at every available opportunity, right? If you give them a consequence-free loop-hole, that will become the game, with the straight and narrow path abandoned as a mere curio.

But in practice, that’s not what happens – or at least, it’s not what happened for me. I didn’t find myself deliberately failing to skip forward, despite really not enjoying the car chases (fussy controls, poor camera when reversing, hard to tell what world objects were solid and which destructible so you’d often try to short-cut through a picket fence only to have your 60mph sports car rebound off it). I didn’t even find myself skipping them after failing. Instead, the sense that, by skipping, I would be missing something over-ruled my desire to move forwards. So I ended up replaying sequences and, eventually besting them. (Except for some, but I don’t remember how many. Which is good too, right?)

What I like about this is that there’s no punishment – the game is totally fair, unbothered even, by what you choose. If you want to care about not playing “properly”, that’s between you and your conscience – and if the problem is the design is badly balanced, that’s an easy call to make. And then, if you do choose to skip, you’re not missing much – you’ve seen the chase (or at least some of it), you know the scenario. You’re not being forced to replay it in a dumbed-down mode; you’re saying “I’m not here for that part, move on, and show me something I’ll like.”

It’s sort-of the linear-game equivalent of a Deus Ex-style multiple-strategy design (as in, sneak around or go in fighting?) And it doesn’t have the same negative effect on the story that replay/easy has: because the sequence was hard, it was a million-to-one-shot; in fact, so much so, you had the stress and tension of the scene because you didn’t overcome it. But the hero of the story did, and the narrative continues unbroken.

(It should be side-noted that this mechanic works in LA Noire because the game-play is quite varied: no car chase is followed by just another car chase.  A similar mechanic in Uncharted or KillZone or Gears wouldn’t fare so well, perhaps, since any gunfight you couldn’t handle might only be followed by another gunfight. Still, there are some sequences where I would have appreciated the option to skip through after being run down by a tank for the twentieth time…)

So all this got me thinking about stuckness in IF. What if every puzzle had with it the option to skip the puzzle entirely – to have the character solve the puzzle for you (or reveal, perhaps, that the puzzle is unsolvable at the time because of dependencies). Would players game this and click the solve button every time? Is this different because there’s no “attempt, fail, repeat” cycle? What about in games where there is a replay cycle, like Make It Good or Varicella? Would it be good to allow players, after a few times through, to simply commit high-level actions like “Search the murder scene thoroughly” or “orchestrate the assassination of the Chancellor using the mounted battery”?

Could a puzzle game measure the amount of attempts the player makes on a puzzle – how many times they examine it, or toy with it, or attempt a failed solution? Would players game that by using the “again” command?

Again, design theory tells me that this is the same as having a WIN GAME command, and that it removes all the challenge, and hence the reward, from the game. But in a content-focused medium like text-based Interactive Fiction – and in a game like LA Noire  - the reward is the content, and the challenge is simply colour applied to that content. And we’re not talking about the option to auto-solve any puzzle first time. The player has to be deemed to have failed, first.

So it’s like hints, but instead of having hints that say – “don’t read the next one, go play the game some more and when you’re still stuck, because you’ve got no new ideas, then come all the way back through the menu system to this point right here and press the NEXT button, which you could press right now, and you’ll get the next hint, which might even prove no more helpful than this one…”  - you have hints that say – “You’ve not tried. Go away and try.” Which is kinda what a teacher does.

To me, it seems like the tricky part here isn’t the concept – I think the concept is a good one. The tricky thing is the metric: how failed is failed? Which is to say, all we’ve done is turn the problem from one of balancing our difficulty well into one of balancing our hints well. But maybe that’s a better problem to have?