I was playing this, and I webbed up an oil tanker that was going explode (this makes it not explode) and Peter goes "geez, if i'd webbed that up a second later...😬" but of course, it would have been fine, they never explode. Nothing bad can happen in this game.

Got me thinking how much more I would enjoy that other game: a version of a spiderman game where you actually have to face consequences for the occasional failure/miscalculation/oversight/fuckup/hard choice. Spiderman movies (and probably comics, whatever, I'm not a nerd) are all concerned at some point with how he is viewed by the community; which is often negatively. I want to have to manage that stuff.

Spiderman 2 has you occasionally end a little "crime reported" sidemission by having to carry someone off to an ambulance; it's the jankiest, least-fleshed-out part of the game, but what if it was a consequence for failure instead of just a random thing? You show up, fight a bunch of guys who want to blow something up, you fuck up, it blows up, and the mission becomes rescuing people from a burning building and salvaging the situation as best you can. You can fail that too, and people are upset with you. You get blamed, and that affects how people talk to you on the street. Your vigilantism becomes a more thankless task, and now when someone actually does cut you a break, it matters.

I'm always swinging past a crime I don't feel like addressing, and I think there should be consequences for that. Why is Spiderman all about being incredibly stressed out, but the Spiderman game is a pure power fantasy? If you're okay with consequences, there's automatic depth: I can't be in two places at once, and two crimes are happening. Make me choose, just by which one I go to. Make it matter how fast I get there. The more I level up and gain skills, or just get better at the game, the more likely I can actually handle both - maybe not perfectly, but I helped!

Which is a staple of those films: juggling too many responsibilities, not always succeeding. In all those films (not just, but especially, the Raimi ones) spidey "just can't win", not just in reconciling multiple identities and schedules but also at every turn in his daily spidermanning; there's always an impossible choice to make. Not here! Everything is neat and manageable, there is no monster you can't put back in its box same-day, and nobody really minds what happened in the meantime. And there's clearly a deliberate choice, and it's a huge bummer, not to deal at all with the "secret identity" issue in these games. Everyone Peter and Miles know pretty much know they're spiderman, and you never have to be out of costume except in really specific handholdy scripted walkntalks.

The entire vibe of the city is off, too: the game is obsessed with New York being amazing, bodega bodega bodega, but the whole thing feels like a cheap hollywood facade. Everyone loves spiderman. Everyone loves New York. Nobody is doing it tough. Nobody's homeless, nobody's desperate. Everything is clean, even the trash is clean. Everyone loves the city, and it comes off ingenuine, not just because of the stilted voice performances and saccarine writing but because the city feels too perfect to love. What are we fighting for? It's an expensive dollhouse, and no matter what I do everyone will love me except J Jonah Jameson, and that's as a joke.

I do actually like this game a lot; I guess I've 87%'d it, and I'll still jump back in to swing around, but the more I think back on it the more I'm disappointed. It's broader and shallower than ever, and I can't imagine how they do another one without thinking quite a lot harder about it.

On Twitter once I did a big thread where I went through Half-Life Alyx's SDK - so this is Source 2 - and had a big ol' yarn about 'em. That thread's turned into a useful resource, so here it is in Cohost Post Form.

Gonna look at Half-Life: Alyx's SDK and tweet about it a little. If you don't know already, I've historically had a big old axe to grind with modern level design tools! Still do. Neither Unity nor Unreal has level design tools worth mentioning built-in. This looks more promising.

Of course, Valve's not Unity or Epic - Valve just makes games; they're not big into licensing. These are internal tools, so there's gonna be jank. It's part of why actually devving on this engine would be a bad idea. But we sure can steal their awesome tool ideas!

An initial interesting thing is that we boot up straight into an Asset Browser, not a level editor. It doesn't assume we want to deal with maps, which is interesting, and reminds me of the days when UnrealEd had an Unrealscript code editor built right in. First thing I'm gonna open is Hammer though.

Already this is hot as hell, UI-wise. We have a lot of uncaptioned buttons that we probably mouseover the first time to find out what it is, and after that it's always one click away - we're not slowing our users down with submenus just so we can fit a caption in. Refreshing!

We've also got geo editing modes front and center, which bodes extremely well. Often level designers aren't able to edit geo in level editors at all, or if they can it's real clumsy. Isn't that wild?? Level design? Sort of important? Gotta iterate? Can't usually do it anymore? Wild. We've also got a vertical toolbar super similar to the old Hammer/Worldcraft one, which was always great. For reference, in Unity and Unreal, if you can reach this stuff at all it's behind many tabs and dropdowns. More clicks, more time, more frustration, less iteration, bad.

Yeah, this is really nice. First time opening the editor! Sure, I'm making trash, but look what I can do! I can drag stuff out and make changes to it super fast, which is great for level design, but also it's more powerful modelling-functionality-wise than something like ProBuilder.

This is the most functionality I've ever seen in a level editing tool, and so far it all operates more intuitively than the same operations in any modeling program I've used (Max, Maya, etc). And I'm barely into this yet!

Here's something I always point out to people making editors: in Hammer (1 and 2), if you select something or drag something out, the dimensions are right there. Nothing else does this. UE4 has a ruler, cool. Why does it not just draw these numbers on whatever I select?

Look at these UV tools. A whole bunch o' buttons for each of Align, Scale, Shift, Rotate, Fit, and crucially Justify. Hammer 1 had these, but this is a better UI. Nobody else bothers with this at all. If they do, there's at least twice as many clicks involved as defeats the purpose.

I'm sure that Very Many of Valve's systems and tools are janky and terrible compared to Unreal's, but these level design tools? I am literally seeing one good design choice and tripping over four others on the way to tweet about it. I am drowning in The Good Shit.

https://youtu.be/pBUddmjob20

Hotspot materials! An incredible feature! It's extremely good! We all would have had stuff like this a decade ago if AAA had even attempted to reconcile increased graphical fidelity with the needs of level design, rather than throwing LD under the Env Art Bus. This little house was made with just hammer geometry and hotspot materials. Looks fuckin' great.


These are all set up in something called the subrect editor, which is dead simple and cool. Like, you can make your own ones of these so easy.

Time to look into something called Tile Meshes which I've also been hearing about for a while. They sound pretty sick.

https://youtu.be/CzCDCPPIKF8

Oh I see, it's sorta like hotspots but for meshes + a bit jankier. You start with a quad and if it's of the dimensions of any of the meshes in the tileset it picks one, also choosing the right mesh for corners. I kind of want the meshes to scale the distance rather than leaving gaps, though.

It's incredibly pleasing that the folks at Valve have even been building their tools from this angle. For years, folks just got artists to churn out dozens and dozens of bespoke and/or modular meshes for mundane shit like this, all it did was cost money and stop LDs from working. These tools are amazing and everything, but none of it is a tech advancement. We could have had this ages ago. We didn't because of people just resolutely not giving a shit about the field of level design, including most people who make in-editor geometry tools. For programmers, modeling tools inside game engines seems to be an interesting challenge for whatever reason; making those tools useful for level design seems not to be.

ProBuilder, for instance, is barely better than nothing for level design, but it's not interested in level design. It just wants to exist for its own sake. And if your in-engine geo tools aren't useful for level design, they're good for nothing at all - artists aren't ever going to use them.

There's also the issue that it's programmers making these tools, and programmers are the people who use tools the least. So the only people able to make these things are the people least qualified to do it. Unless your coders are like, multidisciplinary or interested and empathetic. If you're making level design or geometry tools for game dev? Ape Valve, or make bad tools. Them's your choices at this point.

Here are some fullbright screenshots of Half-Life Alyx maps made largely out of Hammer geo:


And here's some with no props, only geo made in Hammer (no tilemeshes either, so of the stuff that's hidden, not much of it was actually bespoke for the scene). In any other current engine, this would be unfeasible to have a level designer do. Someone whipped up these concrete hallways, with their grates and trim and worn edges, in no time at all, without an artist being involved.




This thread has had eyeballs on it from Valve, Epic and Unity, but I'm not too optimistic about it having an impact - engineers' ability to stare directly at this problem without seeing it has held for 15 years and might easily hold for 15 more. Still, can't hurt!

Someone suggested that this might be less of a problem in Unreal if Epic were still doing linear games, and it's actually interesting that I don't think that would do it. After having had great LD tools in UE1 and 2, Epic actually sort of led the charge away from them around UE3. Their in-editor geo tools got worse from there, and continuing into UE4 as support gradually went away entirely. Things like Gears of War wanted the extra fidelity you could get from making maps out of expensive Lego pieces versus their existing LD tools, and they could afford the pieces. It's a worse workflow, but it "works" if you can afford it.

Of course, it doesn't work enough to hold a candle to the work on display in games that licensed UE1 or 2, and it doesn't work at all if you're not extremely cashed up. And bit by bit, you push your level designers away from their levels, and start designing the game itself around that absence.

Unreal has been the engine to chase for a very long time, nothing has ever seriously competed with it, so these decisions have rippled out. If you used it, which a ton of AAA did, it was make levels the Epic way or make a lot of work for yourself that you might not be equipped for, and the effects of that flow on forever to affect indie development and everything else.

So dogfooding works for a lot of things, and it's a lot of why Unreal is an amazing engine, but you'd need more than that for this. You'd almost have to build some massive project around understanding the things that have been lost. Feels like Hammer 2 was that.

In summary, let level designers make levels and we'll just call you when we want the rest of an owl

It's pretty situational I guess, but I usually enjoy things less the more abstracted or board-game-rulesy they are, especially in a real-time context. But the less that stuff is exposed for what it is, the less I mind it - I probably don't mind having a 10% chance of headshotting a guy, but I do mind knowing about it, versus just taking the shot if it feels like I can make it.

Half-Life 2 does an interesting thing nobody knows about, where bullets fired by NPCs appear to just be sprayed out, but actually are doing a roll on whether they hit the target or not, based on a value that represents the character's aptitude with the weapon they're holding, which is tabled somewhere - eg, cops might be worse at using shotguns than soldiers are. This seems a little overengineered for what HL2 ended up being, and functionally I don't think it really ends up executing (HL2 gun-wielders are pretty much turrets in the end) but it also hooks into another cool system where if a bullet has rolled a miss, it tries instead to hit something interesting near the target, like a physics object.

Only vaguely related to the question, but here are some old thoughts on ways to make hitscan weapons feel cool and fair in an FPS. but the HL2 thing seems in line with this - considering accuracy to be a person value, not a gun value.

separate from throwing it. 90% of the potential interactivity of a grenade is collapsed when it explodes

in a game where you can throw a grenade without pulling the pin

-force enemies to break cover, unaware grenade is not live

-dialogue opportunities when they realise

-a free "throw distraction" verb without adding a system

-shootable non-live grenade doubles as remote bomb

-can throw one in somebody's face, stunning and freakin' em out

-share grenades with pals

-risk/reward, enemies can pick em up. "hey free grenade"

-enemies approach non-exploded thrown grenades tentatively, because throwing a non-live grenade is a weird thing to do. what if it's just defective

-if you do this too much they start expecting it and getting cocky about your grenades, with hilarious consequences

as long as i am only throwing a grenade to flush somebody out of cover, why waste it. foolishness

is it

https://www.youtube.com/watch?v=VVAWi6K0JzY

the jury's still out on the sim part

imo, yeah a little bit? but it's not much of one.

i reckon the core of imsim is "things are systemic", they don't work/not work arbitrarily or case-by-case, and also the world has to be designed as a world, not a level, filled with those systems. this is how you get cool emergent results, and in chaos theory you don't

unlike SC1, CT applies its rules consistently, so it's less frustrating, but the levels are still just levels: there is one way to go, or sometimes a right way and a wrong (backtracking/circuitous) way. it's also full of freestanding systems that don't really connect with each other or with you. the "use thermal goggles to push keypad buttons in reverse order of how warm they still are after a guy just used the keypad" system from SC1 is still there, but there's no point using it. there's a new "toggleable glass" thing where it toggles between opaque and translucent if you zap it, but it comes up one time and might as well not be there. there's the RFID-beacon lasers, but again, one time, in a hallway situation, and it's not like you can trap a guy by EMP-ing his beacon or anything. sticky cameras are still a solution in search of a problem because you can always look at a thing firsthand.

but all this stuff is still built systemically, unusually so, to the point where if they put out level design tools for the single player, i think someone could make a splinter cell game that i would consider an imsim with zero code. so that's interesting

You know how when Deus Ex 3 first came out, it had boss fights that you couldn't get around, you were just forced to fight and kill a guy? And everyone was mad? And later the devs patched in other options, basically one for every style of play you might be into?

Well I've decided they were right the first time

Deus Ex 1 is the big amazing immersive sim nobody can live up to, because it's pretty uncompromisingly systemic. A lot of designers interpret "systemic" as "every problem has several different solutions depending on what kind of player you are", but that's not it. That's just "design five locked door puzzles".

Deus Ex forces you to specialise (to what degree is slightly up to you), you can't be good at everything, and no one skillset gets you through every situation. There's hackable keypads, pickable locks, breakable doors, robots, computers, findable codes and keys and stealthable guys, but at the end of the day, sometimes a big robot is in a narrow hallway and you have to blow it up. If every situation contains an accommodation for every skill - "there has to be a vent to go around the robot, a computer to remotely hack it, a killphrase to yell at it", every single time, that's not a sim, it's a choice of numbered corridors where players commit early on to which number they'll always pick.

Deus Ex is a sim because it's "keypads exist, hacking exists, lockpicks exist, things are breakable, things are heavy, things are poisonous, things are electrified, etc etc" and then it deploys all those systems in contexts which don't care that those systems exist. Which isn't a style of level design that works well for games that aren't sims, and the lack of it is why 3 doesn't land as a sim. In 3, you're never like "hmm, I like to do X, but that's not an option here". It's always an option, because the game cares that your type of player exists more than it cares whether the place you're in could exist. The boss fights were the only exception. They were the smart bit!

Most designers would play a Deus Ex level full of keypad locks and say, "Hmm, they're not giving Lockpick Masters enough to do here. They need a better balance of keypads to conventional locks!" But that's wrong, because it's a sim. The building needs to not care where you've invested your XP. If it cares, the player is never up against anything. The best moments and player stories from immersive sims are things that most designers would have optimised out.

There's a possibly-apocryphal story about Warren Spector insisting that everywhere in Deus Ex 1 be based on a blueprint of a real place, which is supposed to be a nightmare thing to do and whatever, but if true, I really think it's a huge part of why it ends up working. It just makes sense. How do you design a space for a simulation? Maybe you barely design it at all.

Hi I'm Joe and here's what I'd do if someone at Valve put a gun to my head and told me to add a new type of Special Infected to the video game Left 4 Dead

First thing to think about is, as a goal, you need an issue you want to address. For me, in Left 4 Dead, that's players rushing. Whether it's co-op or or versus, players in L4D can pretty much just speedrun everything, if they want, and get away with it. I'm hostile to this. I don't like those players - they can only have a good time if everyone else has a bad time.

So, my Special Infected is the Lurker.

-The Lurker always spawns ahead of the survivors on the escape route (an L4D concept which is just "the path from map start to map end").

-It doesn't spawn very often, and only once per map.

-It looks and acts identically to an idle common infected (normal zombie), and does all the usual idle behaviours: leaning on walls, staggering around, vomiting.

-If any player gets too close, it erupts into a terrifying mess of bloody tentacles like The Thing and does low-damage minor-knockdown slashes within a several-meter radius, still staggering around. It doesn't incapacitate like other special infected, but it knocks you around.

-If it takes too much damage, it disengages and tries to escape by getting low and snaking off ahead. If it gets away, you might encounter it again, back in stealth mode.

-It doesn't do much damage, but it has a lot of health, so fighting it takes time, and draws some attention. It's also faster than a player, so you can't just run away.

-Pre-aggro, there's two ways to know it's not a common infected:

1. Regardless of noise or damage, it never aggros unless you're close to it, it just keeps doing those idle zombie behaviours.
2. If you shoot it in the head, it "dies" like a normal Z, but like 5-10 seconds later it gets up again.

This means you can't confidently detect a Lurker without drawing attention from other enemies by shooting (which matters a lot or a little, contextually), or by slowing down and taking a few seconds to be sure. You're disincentivised from rushing because the level of danger ahead is unclear, since every zombie is a little bit of a suspect, at least some of the time.

I think this could be a pretty good solution to how, for a certain type of player, a lot of the survival-horror tension goes away when you realise that instead of cooperating with the experience, you can just run fast.

In versus, as the Lurker, you get a third-person view, and play the The Ship type game of trying to look plausible as an NPC, while trying to get close to an area where the survivors will have to get close to you. For instance, playing as a Lurker, you might stagger into the path the survivors need to stealth along to avoid disturbing the Witch.

I wish L4D supported modding so I could actually give this a try!

Thanks for coming to my Zed Talk

instead of the DLC giving you access to quite a slow motorcycle, which sort of undermines horses and climbing and ends up not being very fun, it should have given you a bicycle

the bicycle uses your stamina wheel. if you pedal real good you can go up hills faster than running. but going downhill, you can go super fast and your stamina recharges. you can maybe hold onto it with your feet while you paraglide, it's not that heavy. paraglide across the map and land, still on your bicycle, going extra fast. you can do a little bunny hop. you can lean around. you can tie those helium balloons to it.

horses are still worth it because they do better with certain terrain and you don't fuck your own stamina up going up a hill. climbing is still worth it because unlike the motorbike the bicycle doesn't do a bonkers climb. also, a bicycle is still pretty high end tech for botw. plus, imagine, you're holding A or whatever to go fast, link is pedaling his little heart out, he's got those sweat droplets coming off his face. so cute! classic botw

it seems like smaller-to-medium-sized game development often starts from recognising some feasibler-than-it-seems, high-bang-for-buck gimmick or tech thing, with which you can reliably wow people, and then imagining an otherwise small-scope game around that

which i think there are many examples of but i don't want to list any

and how that used to seem smart to me (build your game around a pretty-cool-no-matter-what core element, so you can't go too far wrong) but now seems more like hedging, something done when you know you need to make something but don't feel strongly about what or why

the opposite thing seems better to me when i see it: something built around a core risk, which a team has agreed to take, and not to mitigate, because they do feel strongly about it

but if ever i'm pitching something for investment i usually feel like i have to do the first thing