Fun little story, at Kansasfest 1992 (the Apple II conference that persists today) I was incredibly young but was roaming the dorms at night between various groups hacking on stuff and begging for underage-beers, when I heard some chatter about some people who had developed an interesting 3D game and promptly received over $800KUSD in shareware fees in just a month, enabling the purchase of fast cars. I ran into those people and saw a brief demonstration of Wolf3D on a plasma-screen laptop, it looked mighty fancy.
The next day, I saw Carmack stick his head into a computer lab where and friend and I were hacking on something, and I said "Hey John! I really liked Wraith! I just beat it!" (Wraith was an 8-bit tiled 2D adventure game that he had released). He smiled and said "Oh, I've gotten a bit better since Wraith..."
Of course by that time, he was about to release Doom. Not to belittle the ray-casting problem of determining which 2D tiles to hide and show in those adventure games, but he had indeed gotten a bit better.
For those interested in learning more about DOOM, the book "Masters of Doom" is an entertaining recollection of how id Software started and how John Romero and John Carmack started their careers.
Fabien is a Hacker News darling, I'd highly recommend both the DOOM black book and Wolfenstein black book, I have them and love every page. Give him your support!
Masters of Doom is also required reading for any hacker.
I second that. Romero's autobiography covers Doom's development in detail, well beyond The Masters of Doom. This includes all the games leading up to Doom, which informed their approach. It was a great read.
What a treat it was listening to two legends talk with great passion about software they made 30 years ago.
I absolutely loved their answer to a question regarding better enemy AI, which boiled down to “better enemy AI doesn’t automatically make for a more fun game”. (they put fun and player experience first and aren’t tricked by shiny new toys)
Was also interesting hearing Carmack lament how so few much loved games release their source code (he attributed doom’s longevity partly to the release of its source code)
The original F.E.A.R. always stuck out to me for its AI. Not because the actual AI is very smart - most enemies still are dumb lugs that try to get the player and have basic "is seen, rush down" AI, but because the player gets a lot of feedback for their actions.
The main way to trick a player into thinking an AI is smarter than it actually is? Have their voicelines "cheat" a bit and respond very vocally to things the player does near them; even if the enemy can't see the player, the idea of "oh shit, they heard that" can really enhance the seeming intelligence of the enemies.
Given that F.E.A.R. is a horror themed shooter, that's a big boon.
It's a testament to the game that its AI is still this memorable. Have there been any advances in that area since then? I'm not at home in the singleplayer FPS genre anymore, how do the SP campaigns of the newer Halo or CoD games hold up?
> Was also interesting hearing Carmack lament how so few much loved games release their source code (he attributed doom’s longevity partly to the release of its source code)
I blame middleware. Hell, middleware makes even releasing the mod tools harder, i remember when i worked at some gamedev company and we wanted to release the editor for our game and the company behind some middleware we used wanted us to buy a separate license for it.
Is there any organization that buys up rights to classic games and then releases them as open source, removing/replacing third party dependencies in the process?
They gave a couple examples of the kind of AIs that don’t make games fun. One was some game (real or hypothetical) where you fight a small drone that’s very smart and fast, so it’s hard to shoot at. Another was where the enemy goes off and concocts a sophisticated plan to beat you.
They described how the player wants to feel like the game play is about themselves, not about some ultra smart enemy. So big and dumb enemies tend to work best, even though they don’t use sophisticated AI by today’s standards.
It was like the the Jurassic Park meme: just because we can doesn’t mean we should.
(Also very strong agree on monster infighting, it added so much depth, as a kid I didn’t even know monsters could infight until the level with cyberdemon and arachnotron in the same room)
Or the Deus Ex “ai” which people praised as feeling very intelligent- but was just programmed scripted sequences based on what the player most likely would do.
Yeah, in the end it's all if-this-happens-do-that under the hood. And that's important because game AI must be deterministic. Otherwise reproducing bugs would be impossible.
Some games are so complex that AI on the higher difficulty levels resorts to cheating. I understand the rationale behind it, but it has always rubbed me the wrong way.
Off the top of my head, several roguelikes (e.g. ADOM, Caves of Qud, ToME) and several entries in the TES series (e.g. Daggerfall, Skyrim) support it.
I also remember at some point playing an RPG where you found an ongoing even fight between two armies and you could join one to make the fight go your way, but I don't remember what it was.
Oh, that was so much fun. The best was to get a strong creature in Quake pissed off at the zombies who basically never died, and they would fight forever until the zombie prevailed.
I think I managed to get a Shambler pissed off at some zombies once, there was much shredding.. but the zombies always won in the end while I hid and watched.
Monster Hunter has it, it's pretty epic seeing massive creatures fight each other and you, a tiny human with an oversized weapon, running for your life and / or awaiting your opportunity to kill them and use their corpses to make new weapons/armor.
FWIW there is Xash3D[0] which while not technically open source, it does provide some of the practical benefits of having an open source engine (the issue is that Xash3D uses some HLSDL headers and needs to link against HLSDK to work but HLSDK is not compatible with GPL - the devs have tried to contact Valve for letting them use HLSDK with GPL but Valve never responded). Specifically it has been ported to a bunch of platforms and supports not only Half-Life but also other GoldSource-based games (e.g. i was able to play Gunman Chronicles using it - i have the original CD version but it has DRM that makes it impossible to play under Linux, meanwhile with Xash3D the game plays natively and i can use raw mouse input and widescreen resolutions).
It is but it was a reasonable leap on Quake - Quake had 256 colour palette for the whole game whereas Half-Life had a 256 colour palette per texture, Half-Life got coloured lighting, Quake had game code written in QuakeC whereas Half-Life had game code written in C/C++, Quake had vertex animations but Half-Life had skeletal animations etc.
Definitely the same bones and I'd say that a lot of engine code would be identical, but there were some big changes that aren't trivial to implement in id Tech 2 to get feature parity with Half-Life.
It's also worthwhile to remember where the team honed their craft, as was also mentioned in this session: The early id team worked for a publisher called Softdisk that provided a game subscription where customers received a new game every month. This was basically a way to iterate on the practice and process of game development in one month cycles. The shareware relase of Doom had four months of development, which sounds crazy short by today's standards, but for them it was unusual to have so much time.
Wikipedia says 1 year: official team development started Nov 1992 and the first episode was released Dec 1993. IMHO this is underplaying it given Carmack started on the engine before that and it evolved from Wolfenstein 3D.
Yep, I remember him adding emphasis while reminiscing the Softdisk era and how forcing themselves to push new stuff on a tight schedule may not have been the right conditions to produce anything of quality but it gave them the right conditions to hone their skills.
What I'm uncertain is, whether this was mentioned in the book "Masters of Doom" or Carmack's interview on Lex Fridman.
I remember when, in 1993, my boss bought a $6000 Windows machine at work - to play Doom.
Edit: He had already mentally checked out of this job and was transitioning (already working at) a different company. As a comp-sci computer graphics specialist, he dug Doom at multiple levels. Me - I was not impressed by Doom - probably because I didn't have his background to understand what a feat of engineering it was to get such performance out of a 1992-era CPU.
If you played other games at the time, there wasn’t really anything else like it. Wolfenstein and similar existed, but Doom was a significant advancement. Us lowly console players were many years from anything similar.
I still feel Marathon was at equal level, even surpassed it in several areas. And released maybe a couple of months after Doom.
But it was only known by Macintosh gamers at the time.
In which ways I feel Marathon was better:
- better light effects
- ability to look up and down
- secondary effects on various weapons
- deep story (done through interactions with computer terminals)
- variety in physics based on levels and weapons (e.g. grenade launcher) is affected by physics
- having to deal with oxygen depletion in some levels
- some AI that aids you (defense drones that support your fight against the aliens)
- more variety in level objectives (e.g. prevent humans from being killed by aliens)
- more comprehensive multiplayer (I believe Doom supported 4 players; Marathon supported 8 - Marathon has more types of multiplayer games than just Deathmatch).
I think Jason Jones [0] was probably at an equal level as John Carmack at this point in his career.
Despite it having more or less the same graphics tech it's a very different game. It's slower paced, more atmospheric, and with more story. The story in doom was a single page of text after beating each episode. Marathon had in-game computer terminals where they drip fed the story of the Marathon and its AI that was going "rampant" (interestingly Microsoft borrowed that idea for Halo 4, the first one that Bungie didn't create).
Playing the limited shareware release of Doom the very first time in single player mode was already incredible, but it was the multi-player mode that was mind blowing.
It would be fun to see a graph with sales stats of NE2000 clones around that time. They were the cheapest Ethernet expansions boards at the time, and one of the few that Doom supported.
Jedi Knight: Dark Forces II still used 2D sprite explosions and weapons in 1997. Was Terminator: Future Shock true 3D? I seem remember it used the Daggerfall engine which was 2.5D.
First generation 3D games were kind of a mess IMO. They usually hadn't figured out the controls yet and polygon counts were so low I always thought they looked worse than sprite based games. It isn't a surprise to me 2D-sprites hung around for awhile. Billboard sprites are still a pretty good/cheap choice for symmetrical things like explosions.
XnGine (the engine behind Daggerfall, Terminator: Future Shock, SkyNET, etc - i think the last game to use it was Redguard) was full 3D however Daggerfall relied a lot on sprites for NPCs and such (most likely because it had a ton of NPCs in towns) but the maps were fully 3D and there were 3D objects in them (the screenshots in this tool[0] that show daggerfall maps make it clear). Terminator used sprites for small objects (weapons, pick up items, etc) but enemies were 3D models as was the world. In Redguard almost everything was 3D.
It wasn't just the engineering. The immersion was unprecedented in videogames, which was supported, besides the obvious rendering improvements, by lighting and sound.
Experiencing for example, lights suddenly turning off and monsters appearing and making noises behind the player, was very intense and novel.
(There were also other revolutionary aspects in Doom, notably, multiplayer)
I was at E.A. working on the ill fated 3D0 OS video subsystem when Doom was released. The game was immediately installed on everyone's PCs, "for research", and for the next week or two absolutely zero work occurred at Electronic Arts.
Being an OS developer for the 3D0 was enlightening, primarily because very little of the highly publicized 3D0 technology actually worked. The OS team was expected to fix the hardware bugs, while maintaining all the marketing hype. Plus the executives were early spending the money they expected to earn and were half checked out during the critical "nothing works this cannot be delivered" meetings. It was the cascading cluster fuck of all cluster fucks, as E.A. had no capacity to realize this "ambition", just executives capable of selling the idea.
Interesting! My vague recollection (as a kid at the time) was that the 3DO really did deliver some impressive stuff but just at a very high price. Of course by the time the Playstation was released it was game over for 3do (and Atari, and Sega).
I worked on the PlayStation OS too. Very different everything. I was Sony Computer Entertainment's first software engineer hire, promptly sent to Tokyo to help finish the system.
Professionalism; EA was basically modeled after rock bands as programmer teams back then. Sony was creating a series of international divisions and appointing star people from other divisions with their own new division, dubbed Sony Computer Entertainment of [Country Name], and an entire campus to run. Their attitude was game developers were tough and critical and needed to be won over, while EA's attitude was game developers were gamers who code and can be satisfied by a pizza party.
I think in 1993 you could easily still spend $6k by buying the top end 486/pentium CPU from intel, 8mb ram, CD-ROM, large (for the time) HDD, 17" monitor, etc. It wouldn't have a GPU but the latest tech always cost a big premium (it might be just 50% faster than a good but non-top-end system but cost 2-3x as much).
EDIT - looking at some ads in the December '93 issue of Computerworld [0] there's a Compaq ad saying "the Deskpro with Pentium starts at just $3,199" (both the "starts at" and the "just" make me think you could make it cost a lot more than $3,199, altho you could surely also play Doom on a cheaper computer).
This doesn’t contradict anything I wrote above. It was possible to get much better bang for buck than the absolute most expensive option but I can believe the above story that you could spend someone else’s $6k to splurge on the most expensive PC you could find.
I don't tried to bring a lengthy discussion, it was a play with memories: remembering assembling your own PC and travelling the city choosing the specific components carefully based on price.
I didn't mean to take it in a negative direction either! And I agree with you anyway, my computer at around that time was also much cheaper than the example given (I think an IBM Aptiva on sale from a shop that was trying to get rid of inventory) and played games just fine (at least for the first couple years of its life)!
It was very high - but this was a business purchase. It was EASY to spend $6k for a high-end PC in 1993. Also consider that $6000 is $12000 in 2023 dollars.
When I think about that period in gaming I just trip out.
We are as far away from that period now as that period was from the birth of video games. Tennis for Two, Space Wars, Pong, Super Mario, DOOM, Quake, Half-Life.
Compare that 20-year evolution to the evolution of gaming in the last 20 years. You can see areas where games have gotten vastly better, not just graphically or cinematically, but with level design, mission design, characterization, contextual integration, procedural elements.
But still nowhere near the explosion of the first 20 years as innovation poured in, and all of the main players were scrappy young companies, ready to innovate in an atmosphere where it still meant something to create entirely novel experiences that challenged the player.
You can also very easily see an industry-wide shift in game design after both GTA and Minecraft, which seemed so far apart at the time but really were about a decade apart. What comes next?
As a teenager during that time it feels easy to say that we idealize those periods of youth, but objectively speaking there was never a period in gaming history with such drastic speed of innovation between the late 90s and early 200s.
For sure; if you look at 15, 20 year old games now vs today's, you can see it's more evolution than revolution, in terms of gameplay mechanics, graphics, scale, etc.
Not that it's not impressive; look at the PS5 tech demos (large scale areas that players can go through fast without loading times thanks to fast storage), or the Unreal Engine demos, e.g. the Matrix one you can download to the PS5 or the one where they highlight generative level creation.
But it doesn't feel as revolutionary anymore. Mind you, that may just be fatigue, or "getting used to" things quickly.
I think objectively 2D to 3D was a massive jump. Also the FPS genre which is maybe the most successful gaming genre of all time being invented. Can't even think of a new genre of game that's been created in the last 10 years. I guess the sandbox game Minecraft was a little over 10 years ago, that's one.
Even Minecraft didn't really invent the sandbox, it just polished and popularized it.
A new genre in the last 10 years? Probably nothing really. We're probably past the point of new genres being invented until the technology dramatically changes.
We'll see some wild Genre Fusion though. I'm excited for that. I can't wait to see what happens when we stop just stapling "RPG" to every genre and start to really innovate.
Nanite and Lumen are definitely witchcraft. That tech demo left me in awe at what humanity has been able to accomplish. "More evolution than revolution", what a fitting phrase.
I wonder if we will live to see another such revolution in the wake of mature generative AI tools? If creating a game of the density and caliber of Red Dead Redemption 2 becomes feasible with a modest team of 10?
It was a wild time; as they said, Doom was made in (just) one year; while in a sense the games of back then are comparable to indie games of today, even indie games take longer to build these days.
That said, it's a work/life balance thing too; the Doom team did pretty much a year of grinding, pizza and coffee fueled 14+ hour workdays.
And just six years between that and HL2/Doom 3, the first big games to introduce proper physics. Shooting a box and watch it move, or the pick up a soda can/shipping container and throw it around was mind blowing at the time.
Not to mention the graphics evolution; compare G-man from HL to HL2.
Deus Ex didn't have full "the cube can rotate" physics but it did a lot to move this kind of thing forward. You can totally throw a can of soda at a cop.
And Splinter Cell (2002) had curtains that moved around you as you passed under them. Many games had individual features that were innovative physics-wise, but I still think HL2 and Doom 3 were the big mind blowing ones, both for me personally and the media coverage at the time.
Anyone know if anyone ever asked Carmack about how he felt about Descent? That was the first 3d game I was exposed to that ever felt like a true 3d experience.
Not only they revolutionized games, but their decision to open source their game engines has so many implications that it is hard to even begin quantifying their impact.
And it's not only the FPS games that left a mark in history, the 2D platformers were awesome too.
> their decision to open source their game engines has so many implications that it is hard to even begin quantifying their impact
What are you thinking of?
As far as I recall, the main benefit has been to developer curiosity but I don't know of much impact beyond that. The games that used Id engines like HL or Call of Duty had to license it anyway right and didn't use the open source versions? I don't recall much of a scene around the open source engines unlike the mod scene that used the official game and assets.
Things like Quake2 running on the Quest uses an engine that I think started off with original source but eh, it's not that significant. Q2 was huge to me at the time but I can't muster the enthusiasm to play it in VR. I look at it briefly and I'm done.
There was Quake II RTX created by Nvidia but again, eh. That and Portal RTX did not make a convincing case for ray tracing. They ran like pigs and didn't look terribly good to me.
There are endless Doom source-ports, quite a few Quake source-ports, ioquake3 is still alive and kicking and spawned Open Arena.
I don't think it's hugely mainstream but the Doom and Quake engines have been modded endlessly.
I was down in the Dallas/Fort Worth area in 2008 for BGG (Board Game Geek convention). I tracked down the address to the id office, and got John Carmack to sign my id ball cap (had it custom embroidered). I didn't meet Carmack, but the kind receptionist took my cap back to his office, and he signed it.
Thief 2. There's an incredible amount of fan mission content for it, ranging from relatively simple levels, to one-off-masterpieces, to incredibly elaborate full-game-length mission packs.
Warcraft 3. It shipped with a powerful but approachable map editor, had thousands of custom maps and game modes, and spawned two separate gaming genres (Tower Defense, MOBAs).
Morrowind and Skyrim. Both were solid base games to build on top of, both shipped with a powerful creation kit, and both have an incredible amount of modded content available for them.
Technically, Neverwinter Nights would also qualify, but that was less of a game and more of a DnD dungeon master sandbox.
If you're looking for any unusual recommendations, I'm not the best person to ask! I've only tried the more popular ones, but The Seventh Crystal, Gathering at the Inn, The Inverted Manse are my favorite stand-alone missions.
Part of this comes from its simplicity. Maps are relatively fast to make. DoomBuilder and TrenchBroom have made more 3D maps practical. Duke3D and its engine has a smaller yet continuously active community. There are others.
Agree. I got into duke3d and shadow warrior mapping back in the day. The fact that these games weren't truly 3d mandated that you designed the maps 2d first, ie drawing the layout of the map top-down, and then switching to 3d mode and assign textures to walls and floor/ceiling, as well as adjusting their height. That forced you to approach level design in a specific way, it was easy to get into.
When unreal was released I tried to create some maps for it, but was quickly overwhelmed. You're basically thrown into blender and supposed to construct a 3d world from cubes and other simple 3d shapes. I ended up with I think one and a half somewhat decent levels, but they still felt clearly inferior to any of the official maps.
The complexity of modern FPS completely kills the modding possibility.
I think DOOM3/Half-Life 2 is the last generation of engines that still allow a decent amount of mods (HL2 does have quite a few high quality mods), but everything afterwards is simply too complicated for hobbyists to compete with professionals.
I think Quake/Quake II/Half-Life were the last generation of engines where one person could easily understand the whole engine.
Doom 3/Half-Life 2 are approachable too but require a bit more knowledge.
Unreal Tournament I also has a lot of mods so I'd extend the line to around year 2000.
It is a bit unique as somehow the package includes all of the code of Unreal Gold (or something close) so modders can easily add single player contents. The editor is also far superior than the VB one bundled with Unreal I.
But still, even with the toolset Skyrim is still pretty time consuming to develop large mods. It won't be possible without its dedicated fans -- which is kind of ordinary for RPG but less for modern FPS.
> Is there another game that has that level of dedication?
Quake is more difficult to mod (map building is more complicated and you need to build models rather than sprites) but it seems to have a few active source ports.
There are still high level playes participating in tournaments with prizes.
Also some custom games scene, but it is much slower than in the past. There is some website that tries to collect custom maps: there are hudreds of thousands of them.
StarCeaft spawned: MOBAs, turret defense, even had something similar to Among Us. ..
Modern-day I'd look at Minecraft or even Roblox; mind you, the latter is less modding and more building things in the tools that the game / dev tools give you.
The kinds of things they've added to the game are absolutely incredible, and it feels like every couple of months a new mod comes out that makes a huge impact on the landscape.
It's very cool. I've been thinking about trying to contribute to some of my favorite mods, but I have to start learning this stuff first.
Back story on the weapons in the pic, from Scott Host (creator of Raptor: Call of the Shadows): "Me and Carmack were at the Dallas Renaissance Fair, and we stopped at "Angel Sword" (they make real medieval weapons). Carmack asked "Which sword I should get?" and I said the curvy one. The Axe was 8000 dollars and the sword was 12,000."
I specifically enjoyed when John Carmack commented that because of its programmability, Quake launched a lot of software development careers, whereas the early network capabilities of DOOM launched a lot of IT professional careers.
I'd love to somehow know how responsible, on it's own, DOOM was for the propagation of Ethernet inside buildings, and then facilitated the adoption of the Internet as a result.
It was fun to watch live. Had it on my calendar for a while. I would have liked it more without the moderator I suspect, just a discussion between the Johns.
They're legends for the original Doom games, however Carmack definitely went off the rails with Doom 3. Although I actually like the kind of "break" in the series with a more traditional horror style game. The Doom concepts were all there in Doom 3 but they just ruined the pacing/feel of the combat.
Doom 2016 was only possible because Carmack was pushed out - apparently what his team was already assembling was absolutely terrible and they ousted him and essentially started from scratch, something I am super glad for, because Doom 2016 perfectly captures the same feeling as the OG Doom imo.
Makes me wonder, with Starfield recently, if Todd Howard needs to be ousted from Bethesda so that ESVI won't suck hard, or if their problem is more Zenimax related. I think it's a bit of both - they still desperately want to use their shitty, outdated Creation engine.
Talking about that pair, there are rumors of threads discussing issues at the doom/quake era, conflicts between them (romero being more about gameplay while carmack about rendering engine mastery). If the rumors are true, I'd love to read those stories.
The next day, I saw Carmack stick his head into a computer lab where and friend and I were hacking on something, and I said "Hey John! I really liked Wraith! I just beat it!" (Wraith was an 8-bit tiled 2D adventure game that he had released). He smiled and said "Oh, I've gotten a bit better since Wraith..."
Of course by that time, he was about to release Doom. Not to belittle the ray-casting problem of determining which 2D tiles to hide and show in those adventure games, but he had indeed gotten a bit better.