Hello! I'm Tom. I designed a game called Gunpoint, about rewiring things and punching people, and now I'm working on a new one called Heat Signature, about sneaking aboard randomly generated spaceships. Here's some more info on all the games I've worked on, here's the podcast I do, here are the videos I make on YouTube, here are some of the articles I wrote for PC Gamer, and here are two short stories I wrote for the Machine of Death collections.
Jepp: 1) Please keep critiquing games by building new ones :)...
Chris Kilgariff: Hey, This game needs to be a mobile phone...
Andrew: Just linked the book club to you, boosting your...
What Works And Why is a thing where I dig into the design of a game I like and try to analyse what makes it good, hopefully to learn from it but also because I love this stuff.
A turn-based stealth game with randomly generated levels and no savegames. You have two secret agents with different special abilities, and you choose from offices of varying difficulties and rewards to break into and steal money, equipment and abilities. You break in by carefully peering round corners and doors, ambushing unwitting guards with your tazers, and hacking security devices from a special vision mode.
If you want a better idea of how it plays, I recorded myself going through one mission, and talked through my thinking and how the game works.
As in XCOM, there are lots of interesting considerations to precisely where you should move each team member each turn. But instead of optimising accuracy percentages, you’re trying to avoid being seen at all – and usually succeeding. This makes it feel much cleaner and more satisfying than turn-based gunfights, but no less tense or eventful. Being directly seen is a rare, crisis-level event, but lots can happen between ‘perfect stealth’ and ‘caught at gunpoint’.
The edge of a guard’s vision will cause you to be ‘noticed’, causing the guard to come closer on his next turn. Being seen by a camera will raise the alarm level and bring guards running, but not fail or damage you. Your tazer only knocks guards out temporarily, and needs time to recharge, leaving you vulnerable. These tensions and trade-offs are what you’re worrying about on a moment-to-moment basis.
It says something that the last horrifying, gasp-out-loud moment I had in Invisible Inc was when one of my agents shot and killed someone. I’d set them in overwatch to hopefully take out a drone, but a human guard walked in front of them. I was distraught at a) wasting my only shot on a guard who could have been tazed, and b) raising the alarm level by tripping his heatbeat sensor. Not at the loss of life, obvs.
I think every stealth game I like gives you an unrealistic intel advantage. Deus Ex makes your enemies short sighted. Human Revolution lets you switch to third person to see round corners. Far Cry 4 lets you tag people to see them through walls. I decided to go all the way in Gunpoint and let you see everything at all times.
Invisible Inc’s is subtle but extremely powerful. You can only see areas your agents and hacked cameras can see, but: you can see if those areas are being watched. The vision of enemies, even enemies you don’t know about, shows up as if it were bright red light. That doesn’t always tell you exactly where these unknown enemies are, but it gives you perfect, reliable, hard-and-fast intel about the most crucial thing you could want to know: if I move there, will I be seen? Having that intel, and being able to rely on it, makes Invisible Inc a game of strategy rather than one of guesswork and risk management.
When you can fail at something but still carry on playing, I call the range of states between perfect success and total failure a ‘failure spectrum’: there’s a spectrum of possible outcomes, and screw-ups can move you towards the failure end and recoveries can (sometimes) move you back up towards success. In Invisible Inc, there are a few different failure spectrums that you could break down something like this:
How much you care about each of those things is subjective in some cases, but together they form a huge range of possible outcomes. Every encounter and decision you make as you play is moving you up or down on that spectrum, so you care about them all.
A big failure spectrum is good because a lot of the most emotional moments in a game happen on the cusp of failure. If you were this close to being seen, your escape is exhilarating. But if failure is a ‘game over’ screen, spending a lot of time on the cusp of failure means a lot of ‘game over’ screens. Each one interrupts your immersion and ends your investment in this current run. It pulls you out of the game, and you find yourself in a menu, then at a checkpoint or a savegame. Mentally acclimatising to how much of your story has been lost forces you to disengage from it, and you have to build up all that immersion again from scratch.
If failure isn’t game over, it’s still nail-biting when to come close to it. And when you do slip over the threshold, it’s just another development in the story you’re creating and living through. The challenge switches to one about recovering from your screw-up, which can be tense and exciting in itself. Each level of a failure spectrum ultimately means the game can make you spend more of your time at that exhilarating cusp of failure without introducing frustrating interruptions.
This one’s harder to define precisely because the best moments it creates are always unique to the situation. But basically I mean: all these cool systems link into each other, so that problems in one can be solved with another.
My favourite example is when I was trying to ambush a guard with triple power armour. One layer of armour means you need a weapon with one level of piercing to harm them. Triple means you need three. I have none. But power armour is a device, and devices can be hacked.
I spent three turns using my Parasite program to break down each layer of his power armour, and one more getting Deckard in position to take him out. At last he’s in range, so I have him run out of cover, stand directly behind the guard, and- nothing. The tazer icon is disabled. I check the guard’s info: 1 layer of armour. It must have regenerated.
This is bad. I used all of Deckard’s movement points to get him right up to the unsuspecting guard, so now he’s stranded there. The guard is facing a wall, he will absolutely turn around on his turn. And Deckard’s cloaking device is nowhere near being ready again. I have the hacking power to put another parasite on his armour, but it wouldn’t eat through even one layer until the start of our next turn, by which point it would be too late. I don’t have any other hacking tools, not even the basic Lockpick that breaks 1 firewall.
But hang on – didn’t I see that for sale? We’re on this mission to buy new hacking upgrades, and my other agent Internationale reached the terminal last turn – she found nothing good we could afford. The Lockpick isn’t good, really, but right now it would be a life saver. And if we sold one of my less useful hacking programs, I think we could afford it.
Internationale is miles from the terminal, but if she sprints full pelt, she could just make it to the console this turn – alerting everyone on the way. She sprints, everyone hears the footsteps, but she makes it. Selling our Hunter tool gets us enough for the Lockpick, and she buys it. I switch to hacking mode, use the Lockpick on the guard’s power armour, and it shuts down. At last, Deckard’s tazer icon lights up, and I click it.
Grab, zap, whomp – he’s down.
That was the guard AI, feeding into the item system, feeding into the penetration system, feeding into the hacking system, feeding into the shopping system, feeding into the noise and movement system.
On another level, I might have been able to have Internationale buy her own cloaking unit and sneak up to the guard with a Mark II Buster Chip to over-ride his armour by hand.
Or if she’d been Xu, he could use an EMP fist-needle to reboot it.
Or Internationale could have Buster Chipped a drone with a penetrating gun mount to just blow through it.
Or if the Safe Alarm Daemon had been active, she could have intentionally opened a safe to trip one and cause the guard to investigate that next instead of turning around.
I’ve even heard of someone intentionally knocking out their own agent for two turns just because guards don’t notice prone bodies: if I’d had a flash grenade, I could have KO’d Deckard myself to save him.
It’s in Early Access at the moment. Obviously I like it already, but I would pretty much always advise waiting until the developer calls a game done – I’m only playing now because I need to judge it for the IGF.
Not least because of fucking cocking shit like this shitting shit:
Oh. Heads up, 'Abort Mission' in Invisible Inc doesn't mean that so much as 'kill all agents on the spot and end campaign'.
— Tom Francis (@Pentadact) December 27, 2014
More What Works And Why
Martin E: Based on your tweets I got the impresison that you are/was quite enthused by this game.
If the shitting shit hadn't happened, would you've kept on playing?
Sly: I can confirm that even now the game is loads of fun (and highly functional), plus its on sale! Also the 'Abort Mission' button now includes a warning about what it does.
Skeed: As I didn't see you mention it at "Failure Spectrum/Evasion" part, you didn't know that injured agent can be saved via dragging him/her in elevator? (or forget to mention that)
Martin E: I've now watched the video of the playthrough and have one question.
At the end, couldn't you've just upgraded Deckards inventory for $300 in order to grab the Corporate Intel and then sold that Intel to cover the costs for the inventory upgrade?
Tom Francis: Skeed: I didn't know that, that's cool!
Martin: I've done that in the past, but the game never explicitly says what they're worth, and it seems to be nothing or next to nothing. I'd do it if I wasn't urgently saving up for other character upgrades to be able to use my new toys.
Martin E: Tom: Thanks for the reply.
Things About Metal Gear Solid V (Spoiler-Free) - a post on Tom Francis' blog: […] What Works And Why: Invisible Inc […]
Metal Gear Solid's failure spectrum — ????? ?? ????: […] Listing it like that makes it sound absurd, but I really think this is one of the main reasons I and so many people end up having such a great time. Moving to these messier states creates stories of panic and improvisation, instead of frustrating game-overs. It’s the same reason it works in Invisible Inc [the following is a quote from Tom’s Invisible Inc. piece]: […]
Michael: Game looks great Tom. Been going through 3 of those 20 something tutorials you made this year. Decided to restart due to me screwing it up with other code.
Have you made any videos about why you decided to go from Unity to Gamemaker?
Magnificentophat: I think he explained it in one of his GHGC devlogs.
I, too, regret this already.