The Payday Stealth Problem

I haven’t just been playing Skyrim recently! In fact, I have some deep design thoughts on a different videogame entirely. And guess what? You get to share them, because if ever there’s a place for half-thought out amateur “improvements”, it’s a blog on the internet!

So, Payday 2 – the videogame about sending lots of virtual cops to a farm upstate where they can run around and play all day – has a few heists that can be or must be completed stealthily. These are generally terrible and I hate them, which is a little odd because generally I am quite in favour of stealth games. I love Hitman and Mark Of The Ninja, for example. Why is this?

For starters, I’ll quickly point out I know I’m not alone here. Of my regular Payday buddies, of 8 or so there’s maybe 2 people who actually enjoy the stealth heists; everyone else vastly prefers going loud. A quick browse of forums for the game show the same pattern. So how can this be fixed for the announced but still distant Payday 3?

I think the issues with Stealth heists can be boiled down to three major problems.

1) The engine and netcode

Payday 2 runs on a game engine that wasn’t originally designed for stealth gameplay, and that does show at time. “Lag” effects  from this and the netcode can make it very hard for players to do things like participate in split-second dashes between cover if the host and clients see guards in different places. There’s not too much to discuss here, as while perfect netcode is always impossible (anyone playing from Antarctica to Europe is always going to be laggy!) hopefully Payday 3 will have a better suited engine that alleviates this issue.

2) Player Engagement

This is the problem where a full Payday group is 4 players, but the heists also have to allow for a player being able to complete it solo. Ideally, you want all 4 players to be engaged in something useful at all times, and obviously the solo requirement makes it very tricky to design maps that do that. Often, stealth maps devolve into one person actually going out and doing the objectives while the other 3 hide in a safe spot.

This is addressable by map and objective design and this does happen in Payday 2 already. A good example is in Big Bank, where players must hack a specific, random cubicle computer. A group can use one player at a security room terminal to briefly reveal the correct one, at which point another player in the cubicles can see it and hack it. At this point in the heist, the whole group is occupied: One in the security room, one in the upstairs cubes, one in the downstairs cubes and one on the roof prepping for the next stage. A solo player can’t possibly move quick enough for the reveal to be helpful so must brute force it by hacking every computer until they find the right one, taking much longer.

However, while map design can help, groups I’ve stealthed Big Bank with prefer to only have 2 people do the computer section, leaving 2 players unmasked (unable to do anything but unsuspicious if spotted) to keep an eye on guard movements. This is because of the final point:

3) Binary failstates

By which I mean if you screw up once, the heist is failed. Technically, this isn’t the case always, but in practice it usually is. Getting spotted in some heists like Big Bank isn’t a failure, but the heist goes loud and players equipped for stealth can rarely handle that, in other stealth heists the alarm going off is a game over.

Sometimes, you can recover from being spotted. If a guard sees you and isn’t in front of a camera (cameras alert off both alerted NPCs and dead bodies) you can kill him before he gets a shot off (an unsilenced shot will alert the whole map) and no-one sees you kill him (civilians panic and eventually call the cops, and a civilian seeing a panicking civilian also panics in a nice failure cascade) and no other guard patrols in line of sight before the ~15 seconds it takes to answer the dead guard’s pager and bodybag him (2 pagers at once is very tight timing and failing to answer correctly triggers an alarm) then yes, you can keep going. But you only get 4 pagers so even if you’re a master at unnoticed killing, the fifth guard down will trigger an alarm. I really don’t like this system, it pushes players towards a pure don’t-get-seen stealth gameplay that is counter to how loud heist work and that as I mentioned, the engine isn’t very good at.

The Solution

I have hopefully illuminated some of the issues that make Payday 2 stealth so unsatisfying, if you haven’t played it yourself.

But I do have a potential solution which I think is rather neat. Simply put, in Payday 3 every stealth heist should be on a soft timer before it goes loud (by soft timer I mean it is inevitable, rather than an explicit on-screen countdown or similar).

Essentially, this means that once players start interacting with the map- cutting fences, opening doors, taking loot, killing guards- in time these things will be discovered and the alarm raised. This too can be delayed – eg by killing the discovering guard before he can do anything – but eventually you have so many impacts on the map that it is impossible to delay the alarm further. The goal here is that it’s a much softer failstate: if a mistake is made early then the heist should still be completable but with a lesser reward, and the better and faster you are at stealthing, the bigger your reward.

Let me illustrate with some examples based on a Payday 3 version of stealth Art Gallery.

Art Gallery is a building in loop layout with about 8 guards, one of whom is watching cameras in a security room that also controls security measures- laser detectors and security grilles. It’s nighttime, so there are no civilians inside and the security measures are engaged. The objective is valuable paintings. Here’s a helpful diagram of the PD2 version I found on Steam!


In a perfect run: the crew starts by sneaking someone into the security room, silently killing the guard in there so he doesn’t trigger his pager. The guy in the room uses the cameras to guide the rest of the crew around guard patrols, disabling security measures in their way and reactivating them once they’re past so nothing looks out of place. A guard goes into the security room at one point, but is eliminated before he can notice anything amiss. Working together silently and quickly, the crew gets all the paintings and leaves. Guards have started to notice missing paintings and comrades and start investigating, but it’s too late.

Now, lets look at a run where it goes a little wrong. Lets say once the crew is in the security room, one of the team turns a corner directly into a guard the camera guy didn’t see because he was occupied turning the laser detector off. Startled, the player doesn’t react fast enough to kill the guard before he triggers his pager. This is a problem: backup is on the way and the remaining guards are alerted. The crew speeds up, grabbing paintings and putting down any guards they see before they can start shooting. The pager response team arrives- they’re tougher, better armed, in constant contact with HQ and on high alert. Fairly soon after arriving, they notice evidence of misdeeds: a body, a missing painting etc. HQ is immediately notified, takes remote control of the building security measures and triggers the emergency response. Steel bars are activated to protect each painting – no further ones can be stolen now. The heist has gone loud and crew now has to get out before the cops overwhelm them, using ECMs to get past the security grilles they now can’t open normally.

Or maybe a guard manages to get a gunshot off instead and neighbours call the cops so they show up early. Or maybe one of the other guards wanders into the camera room, notices the operator is missing and starts searching around for him.

You’ll notice the AI is a lot smarter in this than in Payday 2. I envision that much like in Hitman, the guards will notice a lot more things about their environment and have different “states”; eg patrolling normally -something wrong – lots wrong – alerted to intruders. This also means that smart players can use this things to their advantage. I would also advocate that guard patrol routes be semi-random; each guard could end up anywhere in the allowed guard pathing area. The general idea behind this approach is that an element of (hopefully) fair randomness rewards map and mechanic knowledge and communication to keep tabs on guard movements. A pro crew will do each heist as fast as possible and adapt on the fly to problems to delay the alert, maybe using high risk tactics like intentionally tripping a laser to draw guards to one location. There’s a lot of decisions to be made at each stage: do we leave someone with the security room guy so he’s not surprised from behind by a guard walking in to get coffee, or is clearing a room of loot faster the better option? And while teamwork helps a lot, a ninja wizard who’s good and quick at dodging guards should be able to complete this solo

I haven’t really touched on civilians or ECMs in this example. With this philosophy, ECMs will be less useful at preventing alerts (though they can still delay them) and would be more used to bypass tougher security measures. Art Gallery is a fairly basic stealth heist and didn’t really have much in the way of that, but there’s lots of room to design more complex heists. Civilians should be easier to move around while tied up as that is a huge pain right now, but to balance that they should be effectively infinite on maps that have streets- keeping your chaos contained to as small and hidden an area as possible should be a goal. Maybe tied up/dead civvies should have some variation of the PD2 pager system attached to them where you have to get to and answer their phones to avoid suspicion? That’s probably a bad idea, I might just be trying to include the funny lines used to answer pagers somehow; I did like those if not the system! There should probably be a mechanism where it gets harder to maintain stealth the more civilians you murder or abduct, but I’ve got to leave some things for Overkill to work out themselves!

I hope you  reading my ideas, they’re probably not perfect and they’re definitely not comprehensive but if Overkill does adopt a philosophy like this for Payday 3, I think it will turn out to be some tense and engaging gameplay.


Oh, a final minor point for improving stealth in Payday 3 is that there should be a training area where every game system is present for learning mechanics, this lets you test out boundaries and probe mechanisms without losing too much. Single player games have the advantage of letting you reload if you do something dumb but in an online game, trying something new and failing loses a large amount of time and effort. For example, the laser alarms on the doors of Big Bank can be tripped by a thrown loot bag. Most players only discover that the first time after they’ve already worked for 20 minutes to get into the vault!






Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s