This design philosophy seems to corresponds very well to Nintendo / Miyamoto classic philosophy. Pick interesting rules, discover emerging situations, and cherry-pick the best situations arranged in an easy to difficult order so that the player enjoy it. I find very interesting that Jon Blow and Nintendo seems to agree on that despite coming from two different terrains: one very intellectual, and another way more down to earth.
I think he's deeply correct about games designing themselves or design being a process of discovery. The Mandelbrot set example works well because math itself has a game-like, interactive quality.
This is a really cool point to zoom in on the mandelbrodt set! Usually when you zoom in randomly it seems to lose an identifiable structure but here its obviously interesting
The trick is to not zoom in randomly, you can start at a random point, but keep focusing in on the interesting structures. Then at the end you just rerender the zoom linearly.
Reminds me of modern RTS games that seem to want to control what the players can do, rather than just create interesting units and let the game figure itself out
I had very interesting experience. I was watching it late at night and it was super captivating to me, but my eyes started closing around 30:00-35:00 so I just went to sleep. and I had like 3-4 different dreams this night about fractals and fractal-like games/phenomenon xD
"authorship is more about having a keen eye for what's interesting.." somehow i'm reminded of stanley kubrick. he never told actors explicitly what he wanted, but he kept doing takes until it worked, even on experienced actors. he didn't want performances, but something emergent.
7:50 This may be just a technical point, but the creators of the video actually provided the content in the sense that they chose the specific complex coordinates of the point to zoom in, which account for O(2 log(S)) bits of information with S being the scale factor of the zoomed-in image. *Almost* every random point would quickly zoom in inside the set and make the image completely black after a relatively small scale factor.
Or the outside with similar results. The Mandelbrot set does have this huge boundary, though. You may think of it as a convoluted line, but it actually has a fractal dimension of two, making its scope more like that of a surface. That gives you lots of detail to choose from. More importantly, what makes a zoom interesting? It turns out to be to end on a satellite of the set. You choose a path for the first half of the zoom. Then you center on an embedded Julia set. The following quarter will copy the structure of the first half, but with double the symmetry and half the size. This keeps repeating at ever smaller intervals until you’re looking at something approaching concentric circles, and suddenly a mini Mandelbrot appears. These points are among the hardest to render with their tremendous iteration density, signifying that something interesting is happening there. There are infinitely many of them. Everywhere on the boundary, you can find one that’s arbitrarily close. Mathematically, these are periodic points, where z cycles back to its initial value. You can locate a nearby one algorithmically using numerical root finding. It’s not a guess, not a coincidence arising from an uninformed choice. There are some other classes with different properties that are less aesthetically sensible but are still interesting. If you just recognize these sample spaces emerging from the structure of the set-a form of truth-instead of presupposing that you want a disk or whatever, you won’t have that problem. Any artistic choice is in picking such a point that you like, with a rendering style that you like. Everything else is truth.
- system / puzzle games - simple rules lead to complex outcomes - player discovers the "truths" and "beauty" (mathematical/physical) - designer's job is not to lead all players through the same exact contrived story/experience/lessons, - rather, to create a statistically large sample/problem space, and - sample/explore its permutations in advance, like a pioneer/leader - by making art for your self first, being honest/true to your self like a diary, you are doing the best possible thing for the audience - then carefully curate your discoveries for posterity - mostly by removing the noise - then by highlighting the beauty - while leaving plenty of space for players to explore for themselves, to find their own beautiful truths - and a way for them to share finds with the community - this can include sandbox game elements, where - players create the content, or - otherwise contribute to the entropy - in this way, these games design themselves - developer's job then is to design/discover/refine the highest yield between - the simplest systems, and - the most emergent properties/outcomes/gameplay
"There are an infinitude of points..." lmao my latin teacher used to put odd endings onto words to see if we were paying attention. Words like "staplize" as in the "the papers were staplized"
That's lovely. Always love when teachers do things that grab students' attention :D That said, I googled it, and apparently "infinitude" is a real word, meaning "the state or quality of being infinite or having no limit". For example, "the infinitude of the universe". Very interesting!
this is all about high-concept puzzle games and how to design them. Any other game has a picture in place, and mostly needs tricks or depth to generate surprises
There is an interesting dichotomy here between thinking about design philosophy in broadly "left-wing", artificial terms, and "right-wing", natural terms. (As broken as "left" and "right" are as concepts. No this is not an endorsement of any political party or figure, and you are not required to change your political affiliations based on your preferred design style.)
I think Blow is actually wrong when it comes to 'infinite content', because just like the fractal zooming, you can procedurally generate content. The real problem is, it would most likely loose it's originality fairly quickly, because it would simply be a tweak of a previous thing already generated. It would be fairly difficult to really go the route of true procedural generation that also affects and changes gameplay as a whole. There's a danger to things becoming too random. I think that's why the planet generation stuff in games like No Man's Sky doesn't really work for a long-term experience. In fact, the initial impression it makes it more 'gosh this looks random' whereas the next planet feels like 'this actually looks familiar but in a different colour scheme'. I think we have yet to really 'break the code' for proper procedural generation that provides mature results, but I do believe it is possible. We can generate cities with rivers, bridges, skyscrapers in places that would make sense and small house on the outskirts of the city. Honestly, I don't think that type of stuff has been really investigated in gaming just yet.
Fair points but I think even existing solutions can work in the right context. For example if NMS was massively multiplayer its procedural universe wouldve been quite serviceable. Its only because the game tries to stand on it alone that it falls short.
You've misunderstood the talk. It is not about procedural generation, it's about a design methodology. The point was that a designer can develop content by further and further investigating of the consequences of the rules of his game, the same way that you can see more and more complexity in a fractal pattern by further investigating the consequences of a simple equation. The point was not to say that games should try to implement a fractal-like procedure in software. It's advice for what designers should do, not what games should do.
escapism game what is going on?!? This is not right, where are the swears and Nazi name calling?!?!? But yeah, both very good points. Interesting talk.
I think that one is for the universe as a whole since space expands faster than light and the universe has only been alive for 16 billion yearsish, so the visible universe can only go up to twice that
@@Dorumin The universe as a whole is infinite we have no size for it. The reason the observable universe is larger than 16 billion ly in radius is because of early universe expansion. Blow just made the same common mistake you did.
@@canadiankdog I don't know, that just feels off, especially when people throw numbers like "infinite" for space and "zero" for black holes. I've always been certain that the universe "edge", or at least as far as matters gets, can't expand faster than light. However, two distant objects in the same universe can get further away from light due to the space between them expanding. This wouldn't have been any different in the early universe, just on a smaller scale and with a lot of more energy in a smaller space to counteract those effects. Never faster than light though
@@Dorumin This is not quite correct. According to our current cosmological model the universe is approximately 13.8 billion years old, and the diameter of the observable universe is approximately 93 billion light years. The reason the diameter is larger than you would expect is because the universe is expanding. For example, consider the observation of a distant galaxy. Photons are emitted by said galaxy and collected by telescopes on Earth (or in orbit). During the time it took for those photons to reach our telescopes, the universe has expanded, and the distance between us and the observed galaxy has increased. We know this because if you perform a spectroscopic analysis of the light collected from these distant objects, you will see that the absorption and/or emission lines are redshifted. Additionally, we determined that the expansion of the universe is accelerating because fainter (more distant) objects were increasingly redshifted. Therefore, we do not know how large the entire universe is because we can't "see" beyond ~46.5 billion light years in any direction. The universe appears to be flat, so it may be infinite; however, we do not really have any evidence to prove that it is infinite in size. Additionally, there may be matter outside of our observable universe, but we cannot be certain because we cannot, by definition, observe it. Remember, the observable universe is exactly that: it is the part of the universe that we can actually observe (measure). As an addendum, black holes do not have zero size. When you perform the calculation using General Relativity (GR), the singularity at the center of a black hole has infinite density (zero size), but the singularity does not actually have infinite density. The reason you get the infinity is because the theory breaks down due to GR being a classical theory of gravity. Since the matter inside a singularity has been extremely condensed, quantum mechanics (QM) becomes extremely relevant. However, because there is no unified theory for QM and gravity, we do not know what is really happening inside the singularity of a black hole. In other words, we do not know the size of the singularity inside a black hole. Everything I've said is obviously not comprehensive. I tried very hard to not get too "into the weeds" and keep things relatively brief.
"The universe has an unlimited supply of generosity and surprise, and we as designers only need to keep our eyes open to what is here." Thanks Jon
One of Blow's greatest talks, a talk I come back to regularly, and it's only been viewed 17k times.
This is his greatest talk ever. Hands down.
45k now. May 2023.
This design philosophy seems to corresponds very well to Nintendo / Miyamoto classic philosophy.
Pick interesting rules, discover emerging situations, and cherry-pick the best situations arranged in an easy to difficult order so that the player enjoy it.
I find very interesting that Jon Blow and Nintendo seems to agree on that despite coming from two different terrains: one very intellectual, and another way more down to earth.
that intro seriously went on for like 4 minutes lol
I think he's deeply correct about games designing themselves or design being a process of discovery. The Mandelbrot set example works well because math itself has a game-like, interactive quality.
Jon Blow, the buddhist game dev.
Carmack is Jesus?
@@jesuschrist7037 carmack isn't a game designer
@@leonardocaetano6307 he said "buddhist game dev", not "buddhist game designer"
Is he really buddhist though? I don't think so.
I think he just believe in the unity of being
I love this man so much
This is a really cool point to zoom in on the mandelbrodt set! Usually when you zoom in randomly it seems to lose an identifiable structure but here its obviously interesting
The trick is to not zoom in randomly, you can start at a random point, but keep focusing in on the interesting structures. Then at the end you just rerender the zoom linearly.
A very interesting take on game design... be open to the interesting patterns in that already exist in the universe and discover them.
Reminds me of modern RTS games that seem to want to control what the players can do, rather than just create interesting units and let the game figure itself out
I had very interesting experience.
I was watching it late at night and it was super captivating to me, but my eyes started closing around 30:00-35:00 so I just went to sleep.
and I had like 3-4 different dreams this night about fractals and fractal-like games/phenomenon xD
"If you answer the questions yourself then you aren't letting the system express itself and it may end up a little bit dead"
I havent even started the video yet but I wanna say the into this video is absolutely amazing -- good music choice too :D
27:30 -> “…when you have eliminated the impossible, whatever remains, however improbable, must be the truth…” - Sherlock Holmes
That was DEEP.
"authorship is more about having a keen eye for what's interesting.." somehow i'm reminded of stanley kubrick. he never told actors explicitly what he wanted, but he kept doing takes until it worked, even on experienced actors. he didn't want performances, but something emergent.
7:50 This may be just a technical point, but the creators of the video actually provided the content in the sense that they chose the specific complex coordinates of the point to zoom in, which account for O(2 log(S)) bits of information with S being the scale factor of the zoomed-in image. *Almost* every random point would quickly zoom in inside the set and make the image completely black after a relatively small scale factor.
Or the outside with similar results. The Mandelbrot set does have this huge boundary, though. You may think of it as a convoluted line, but it actually has a fractal dimension of two, making its scope more like that of a surface. That gives you lots of detail to choose from.
More importantly, what makes a zoom interesting? It turns out to be to end on a satellite of the set. You choose a path for the first half of the zoom. Then you center on an embedded Julia set. The following quarter will copy the structure of the first half, but with double the symmetry and half the size. This keeps repeating at ever smaller intervals until you’re looking at something approaching concentric circles, and suddenly a mini Mandelbrot appears.
These points are among the hardest to render with their tremendous iteration density, signifying that something interesting is happening there. There are infinitely many of them. Everywhere on the boundary, you can find one that’s arbitrarily close. Mathematically, these are periodic points, where z cycles back to its initial value. You can locate a nearby one algorithmically using numerical root finding. It’s not a guess, not a coincidence arising from an uninformed choice.
There are some other classes with different properties that are less aesthetically sensible but are still interesting. If you just recognize these sample spaces emerging from the structure of the set-a form of truth-instead of presupposing that you want a disk or whatever, you won’t have that problem. Any artistic choice is in picking such a point that you like, with a rendering style that you like. Everything else is truth.
The point is, there's all of this complexity that just exist out there in the "mathematical world", waiting to be explored.
11 years old now
- system / puzzle games
- simple rules lead to complex outcomes
- player discovers the "truths" and "beauty" (mathematical/physical)
- designer's job is not to lead all players through the same exact contrived story/experience/lessons,
- rather, to create a statistically large sample/problem space, and
- sample/explore its permutations in advance, like a pioneer/leader
- by making art for your self first, being honest/true to your self like a diary, you are doing the best possible thing for the audience
- then carefully curate your discoveries for posterity
- mostly by removing the noise
- then by highlighting the beauty
- while leaving plenty of space for players to explore for themselves,
to find their own beautiful truths
- and a way for them to share finds with the community
- this can include sandbox game elements, where
- players create the content, or
- otherwise contribute to the entropy
- in this way, these games design themselves
- developer's job then is to design/discover/refine the highest yield between
- the simplest systems, and
- the most emergent properties/outcomes/gameplay
"There are an infinitude of points..." lmao my latin teacher used to put odd endings onto words to see if we were paying attention. Words like "staplize" as in the "the papers were staplized"
That's lovely. Always love when teachers do things that grab students' attention :D
That said, I googled it, and apparently "infinitude" is a real word, meaning "the state or quality of being infinite or having no limit". For example, "the infinitude of the universe". Very interesting!
How about buglarize?
Brilliant
This talk hits different after TotK
Why would you say that ?
It's amazing he already was deep in designing The Witness and building out the team already in 2011? Or is that date wrong?
The date is correct. He started working with a team in 2009.
The man is a machine!
7:16 and that's quite a trip Kappa
4:22 to skip intro
Thanks dude, 4 minute intro wtf
I would say that it is important to watch it if you want to better appreciate what he is talking about... But hey thats just me.
Yeah but not four minutes worth . . . thanks for sharing the talk but to be honest didn't find it that interesting but that's more to do with Jon Blow
Even though I did speed the video up by the end of the intro, I think it does give an interesting perspective on what he is about to talk. :)
At 4 minutes length... I think you should understand that's not actually an intro LOL. I liked it.
💛💚
this is all about high-concept puzzle games and how to design them. Any other game has a picture in place, and mostly needs tricks or depth to generate surprises
This explains the Artifact Card game failing
32:25 these puzzles are triggering my ocd cause they're both unsolvable
The second one is solvable I think
@@nin10dorox oh yeah, I wonder what I was smoking when I made this comment 🤷
Haha, came to the comments to look for this. 😂
There is an interesting dichotomy here between thinking about design philosophy in broadly "left-wing", artificial terms, and "right-wing", natural terms. (As broken as "left" and "right" are as concepts. No this is not an endorsement of any political party or figure, and you are not required to change your political affiliations based on your preferred design style.)
too much question cant ask
I think Blow is actually wrong when it comes to 'infinite content', because just like the fractal zooming, you can procedurally generate content. The real problem is, it would most likely loose it's originality fairly quickly, because it would simply be a tweak of a previous thing already generated. It would be fairly difficult to really go the route of true procedural generation that also affects and changes gameplay as a whole. There's a danger to things becoming too random. I think that's why the planet generation stuff in games like No Man's Sky doesn't really work for a long-term experience. In fact, the initial impression it makes it more 'gosh this looks random' whereas the next planet feels like 'this actually looks familiar but in a different colour scheme'. I think we have yet to really 'break the code' for proper procedural generation that provides mature results, but I do believe it is possible. We can generate cities with rivers, bridges, skyscrapers in places that would make sense and small house on the outskirts of the city. Honestly, I don't think that type of stuff has been really investigated in gaming just yet.
Very intelligent comment.
Fair points but I think even existing solutions can work in the right context. For example if NMS was massively multiplayer its procedural universe wouldve been quite serviceable. Its only because the game tries to stand on it alone that it falls short.
You've misunderstood the talk. It is not about procedural generation, it's about a design methodology. The point was that a designer can develop content by further and further investigating of the consequences of the rules of his game, the same way that you can see more and more complexity in a fractal pattern by further investigating the consequences of a simple equation. The point was not to say that games should try to implement a fractal-like procedure in software. It's advice for what designers should do, not what games should do.
@McN Also a very intelligent comment.
escapism game what is going on?!? This is not right, where are the swears and Nazi name calling?!?!?
But yeah, both very good points.
Interesting talk.
i believe i got it
4:16
I don't know where he gets his numbers from, but the number I know for the diameter of the observable universe is about 90 billion light years.
I think that one is for the universe as a whole since space expands faster than light and the universe has only been alive for 16 billion yearsish, so the visible universe can only go up to twice that
@@Dorumin The universe as a whole is infinite we have no size for it. The reason the observable universe is larger than 16 billion ly in radius is because of early universe expansion. Blow just made the same common mistake you did.
@@canadiankdog I don't know, that just feels off, especially when people throw numbers like "infinite" for space and "zero" for black holes.
I've always been certain that the universe "edge", or at least as far as matters gets, can't expand faster than light. However, two distant objects in the same universe can get further away from light due to the space between them expanding. This wouldn't have been any different in the early universe, just on a smaller scale and with a lot of more energy in a smaller space to counteract those effects. Never faster than light though
@@Dorumin K maybe there's an edge but that's not where that number comes from.
@@Dorumin This is not quite correct. According to our current cosmological model the universe is approximately 13.8 billion years old, and the diameter of the observable universe is approximately 93 billion light years. The reason the diameter is larger than you would expect is because the universe is expanding. For example, consider the observation of a distant galaxy. Photons are emitted by said galaxy and collected by telescopes on Earth (or in orbit). During the time it took for those photons to reach our telescopes, the universe has expanded, and the distance between us and the observed galaxy has increased. We know this because if you perform a spectroscopic analysis of the light collected from these distant objects, you will see that the absorption and/or emission lines are redshifted. Additionally, we determined that the expansion of the universe is accelerating because fainter (more distant) objects were increasingly redshifted.
Therefore, we do not know how large the entire universe is because we can't "see" beyond ~46.5 billion light years in any direction. The universe appears to be flat, so it may be infinite; however, we do not really have any evidence to prove that it is infinite in size. Additionally, there may be matter outside of our observable universe, but we cannot be certain because we cannot, by definition, observe it. Remember, the observable universe is exactly that: it is the part of the universe that we can actually observe (measure).
As an addendum, black holes do not have zero size. When you perform the calculation using General Relativity (GR), the singularity at the center of a black hole has infinite density (zero size), but the singularity does not actually have infinite density. The reason you get the infinity is because the theory breaks down due to GR being a classical theory of gravity. Since the matter inside a singularity has been extremely condensed, quantum mechanics (QM) becomes extremely relevant. However, because there is no unified theory for QM and gravity, we do not know what is really happening inside the singularity of a black hole. In other words, we do not know the size of the singularity inside a black hole.
Everything I've said is obviously not comprehensive. I tried very hard to not get too "into the weeds" and keep things relatively brief.
"I don't like Japanese games either"
Why do I have to sit through 3+ minute unessecary intro? Sounds like something john blow would complain about.
And btw i get the point of the intro, i know what the madelbeot set is, a 30 second to 1 minute intro wouldve sufficed
4:22
@@Bronco541He specifically mentioned the reason for the length of intro.
Also, remember the medium lol. This is a conference.