Doom modding: 2035

Just a fun little speculative thought I had while Skyping with 40oz on his podcast...I think what could make it possible to mod modern Doom would be the ability to act as director and basically engage an AI in conversation. You would talk back and forth while it slowly constructs every element of your map. This would make everything a breeze: new monsters, music, textures, complex layouts and AI pathways could all be dictated. Now, this will take many years of research and development, but I think that by the 2030s we should be seeing examples of intuitive design via humans communicating with AI. It would be like having your own modding team. Mistakes could also be accounted for; you will be able to tweak everything, both broadly and individually. I think this is probably the only way it would be possible to mod games in the future because of how complex they are becoming.

Those dirty tincans taking all the jobs.. Even the ones with no payment!

An interesting thought, though I'd like to note that it is a shame that with every new convenience created the value of the artwork made is cheapened. Would a Mech map be so special if literally anybody could make one by telling their computer "I want a complex, intertwining layout with few orthogonal lines and grand visuals?" I wonder if copies of paintings were worth more back when they had to be hand-painted, or if music was more meaningful when recordings weren't possible... Should we have more respect for a singer who sounds great without autotune than one who sounds equally great, but only with it? Will these maps be claim-able by their authors or will credit need to be given to the person who made the program that actually built the map?

Thankfully this topic in Doom mapping/modding is at least 5-10 years away.

Here's the thing. If you're a crappy visionary, your art will be crappy, too. Just because the AI knows what you want, doesn't mean what you want is aesthetically or mechanically pleasing. People who have an artistic eye, who understand layout and pacing, will still be the best mappers.

AKA: Not going to happen.

We have functional flying cars in 2017, they're all over YouTube. They're just not consumer friendly. And I think arguments like this negate all the times we underestimated technological progression. Recall that Bill Gates once proclaimed that no one needs more than 640k of RAM.

In 2035 I want to be able to mod Doom on a flying scooter.

1 hour ago, GoatLord said:

Here's the thing. If you're a crappy visionary, your art will be crappy, too. Just because the AI knows what you want, doesn't mean what you want is aesthetically or mechanically pleasing. People who have an artistic eye, who understand layout and pacing, will still be the best mappers.

Thats not true; any ole idjut can understand what makes a good Doom map if they simply actually play maps from the community. How many good/bad/ugly maps does one need to play before they can understand the basic standards? Anything after that is purely their vision for their artwork based on their own preference, a la minimal detailing vs Torm-style super-trims or slaughter/challenge arenas vs incidental combat. Most newer mappers don't make cool stuff because 1) they don't know how to yet and 2) they aren't aware of the standards yet from having not played enough current pWads, but not because they lack some secret eyeball passed down from generation to generation when the boy/girl is on the cusp of (wo)manhood.

Doom mapping is as much analytical as it is creative. Much as with writing, you never submit your first draft; you tweak what you do have and make it better. This is one of the main facets of the artwork that is Doom mapping, to me at least.

Doom mapping is a skill, not a talent, but a computer taking away that skill's function/purpose will cheapen the art's worth, or at least that's my take on it.

In 2035, I won't need to worry.I will no longer be of this earth.Damn, I'll miss all the fun!

I had a response here, and now I don't like it. I'm not sure I agree that Doom mapping is pure skill. I think that innate visual/spatial/mechanical aesthetic plays a part here. Ever play a professional, well-paced map that felt kinda lifeless?

57 minutes ago, GoatLord said:

I'm not sure I agree that Doom mapping is pure skill. I think that innate visual/spatial/mechanical aesthetic plays a part here. Ever play a professional, well-paced map that felt kinda lifeless?

Yes, I'd have to agree. DooM mapping is part mechanical manipulation of vertices, part patience and dedication, part strategic thinking, part imagination, and part artistic vision. Just the proper selection of textures requires a good eye, as does the creation of angles and detail.

But getting back to the original post, I'd say the notion of a computer bringing a modder's vision to life is an interesting one. I can easily see this becoming a reality in the next 18-20 years. However, as Fonze pointed out, if everyone will be able to do it, then how special will it be? My guess is that there will be some who are better a having their visions realized (or, indeed, have better vision); those are the people whose work will be in greater demand.

@GoatLord's reply:

Yes, and there is certainly some internal vision that goes into making an interesting map, but do not mistake internal for innate. Were you born knowing Doom? I was born knowing only life. Does a genetic predisposition trump a learned/practiced/honed skill, or is there a point at which "if you dont use it, you lose it?"

I believe genetic predispositions (which is about the more accurate term for "talent," especially for clarity's sake) will take us as duffers to a certain point that might be further than a duffer without that 'talent,' but at some point without practice your 'talent' will remain crude and unpolished. Once the world of 'skill' becomes involved it's more about work-ethic and want/need to improve than raw ability.

Further, and let's switch over to an analogy here for a minute: there is no such thing as a "mapping" or even a "playing guitar" genetic predisposition directly. Maybe your fingers are longer, so can hit those stretches easier (which helps a lot in the beginning), maybe your finger muscles are stronger and can hit the notes cleaner in the beginning, or maybe they contain more fast-twitch muscles and so quick licks/chord changes are easier. Then again, maybe it's just a better hand-eye coordination. Maybe you're just better at recognizing patterns, so the learning part is easier.

Point is, there are millions of tiny facets that go into these large activities that we call "mapping," "painting," or "playing music." Humans didnt have Doom thousands of years ago; where did that talent come from exactly? What is the quantifiable difference that separates one with talent and one without but those tiny facets, which all add up in the end, sure, but much as we can work on our memory skills, we can work on our pattern recognition, our finger strength, our ear, our chops, etc. which goes back to the question what really is a talent/genetic predisposition but an excuse for why you I didn't get there?

5 hours ago, Fonze said:

An interesting thought, though I'd like to note that it is a shame that with every new convenience created the value of the artwork made is cheapened. Would a Mech map be so special if literally anybody could make one by telling their computer "I want a complex, intertwining layout with few orthogonal lines and grand visuals?" I wonder if copies of paintings were worth more back when they had to be hand-painted, or if music was more meaningful when recordings weren't possible... Should we have more respect for a singer who sounds great without autotune than one who sounds equally great, but only with it? Will these maps be claim-able by their authors or will credit need to be given to the person who made the program that actually built the map?

 

Thankfully this topic in Doom mapping/modding is at least 5-10 years away.

I think you can easily envision people in 1997 having this conversation about mapping 15+ years off in the future. Because the tools available today are obvious way better and faster than what was available back then. What happens is that the standard rose. If the 'typical map' is on par with current-Mechadon's stuff (but smaller, lol), 20 years from now, then future-Mechadon's stuff (or a comparable person with that talent willing to put in that amount of effort) will be even more amazing. I think this is generally going to be true, ad infinitum, until the AI reaches a point where it's so strong that you can just say 'make a great map with [suite of dozens of specific qualities that can be specified and tuned] with [random seed so that all such maps with that specific set of qualities don't look identical]' and have it be made without any effort on your part, and have it look human and original, instead of just 'ultra competent OBLIGE stuff'. And I think that, or even being able to make a 'complex, intertwining layout with few orthogonal lines and grand visuals' (again, and having it look human and original instead of just ultra competent OBLIGE stuff), is probably a lot more like 100 years off than just 5-10. I think you are underestimating the 'artistic eye' part; I can see it being somewhat easy to create very competent but ultimately tasteless maps with the sort of AI discussed in the OP, and the mapper's role would be in putting what the AI is capable of doing into something artful.

Lol yeah, I'm sure that people had this exact conversation then.

Wrt the artistic eye part: I agree 100%, and perhaps my number of 5-10 is a bit off (for our small Doom community), but then again computer AI is taking large strides; it's not unlikely that we will have computers capable of all that and more sooner than we may expect, just comes down to how smart/inventive the programmers are as our technology gets exponentially more powerful. But the question of when is perhaps a fruitless one to ask, and as for our small community, though our programmers are both smart and inventive, they both cannot do this forever and might have a different goal to work on with their free/Doom time, so maybe 100 years, heh. I think you worded it all right and I don't discount that artistic eye in terms of "as technology approaches the end point of all this," I was more thinking end picture.

On the note of ever-increasing standards: I also agree 100%; funny that you brought up mappers in '97 as I was thinking about that too with regard to the whole 'cheapening' thing; what was acceptable is no longer the standard now, but then again we have better tools today then they had back then, as you said. One nice thing about our community is that people generally take the time to note when maps were made, looking at 90's maps differently than maps built today, for example. Of course this has all been done without computers actually making the maps for us so that standard will just forever increase, as you said, though I'd suspect that each influential new tool we get (map editor, map format, source port, editor plugin, etc.) has a larger impact on those standards increasing than individual mapsets. Still, nothing wrong with increasing the standards (and the modern Doom SP community is built around the constant barrage of new maps, so it's not like maps won't be 'left behind,' or whatever better word there is, at some point anyway); just means better quality maps for us to make and play in the long run :)

Improvement is always a good thing; hopefully my words before didn't come across as me being against it.

Fonze, I was thinking about genetic predispositions and how much that sort of thing really plays a part in mapping ability. I suppose one could look at it this way, using an example from my own life. For whatever reason, I, person A, have a natural talent for drumming, which was augmented by taking professional lessons for years. Someone else, person B, who is not naturally good at drumming, but who is passionate, could potentially take lessons as I did, and practice frequently as I do. But at the end of the day, I would still have a natural advantage, because I have the capacity to be spontaneously creative with drum patterns and solos, in a way that person B would not. It's something innate, unlearned, that I possess.

Doom mapping is improved not only by having a natural sense of aesthetic, but also a natural sense of design. Now obviously "design" covers a lot of ground, but Doom's design revolves around spatial relationships, flow, pacing, the placement of game elements, the creation of combat scenarios, and the capacity to create organized pathways that engage the player. There's no reason there wouldn't be something inside, something within the soul, that is just naturally good at dealing with these elements of design. Think about how many Doom clones there were, and how most of them were forgotten because they weren't designed well. It indicates to me that the people who worked on those games were not as innately talented with design as Doom's team.

I think Fonze raises a good point. It certainly will cheapen the artistic value of mapping. Mechadon's Vela Pax is an excellent example. If the AI is easy to manage, the skills in making a huge cool map are reduced to intuition and visualization and not necessary the work of sharpening your craft and the discipline to work on it for however long it takes to make every part of it great. It may even potentially make Doom less fun and interesting to play.

However, I dont believe this is any grounds for this future to be unlikely. I'm sure the invention of the automobile made a lot of people excited, and another group to feel that a gas powered engine attached to a chair would trivialize the adventure of travel, and devalue the skills associated with horseback riding. They were certainly right. Everyone is expected to own a car now, and sitting in traffic blows. But this is where the future went.

What we're talking about here are different levels of sophistication within this imaginary AI tech. The first versions will be pretty sloppy, and still require quite a lot of hand-holding. By the time we get to 2035 it will probably be pretty good, at least good enough to design by engaging in very thorough conversation.

But eventually, it'll just be you having a thought, even a vague notion, and the AI does the rest. That's where it could possibly get really ho-hum, where everything looks similarly gorgeous and plays really well because the AI has gotten so good at intuiting what humans want. I think when we get to that point, mappers will have to get really inventive to stand out. The lazier ones will just tell the AI they want something cool and well-designed whereas the deep thinkers will be virtually sculpting far more specific, ambitious layouts, with their own unique look and feel.

Basically, the quality of maps will, by then, be largely determined by the willingness of the mapper to really plan things out, to approach design in a compelling rather than procedural manner. I think this could possibly result in a more coherent and satisfying experience for mapping teams.

Reminds me of that mindflex game where you a headset measured brain activity to levitate and guide a ball. The less focused you are, the quicker you'll have a 1994 map with untextured walls and confusing layouts.

btw this game came out in 2009.