Watching this 6 years later, I can't help but think of Star Citizen. It certainly won't be on the same level as Gravity's Rainbow, but I think that the fact that there is constant backlash around its development time demonstrates that most gamers do not even support the idea of an ambitious 20+ year project, and certainly no publisher does either. Maybe the saddest thing about the 21st century as a whole is that we've lost the ambition to do big things... it may not be in everyone's department to work on 20yr+ projects, but damn, that department should be the one with the most funding and respect instead of the least funded and the least respected.
I had an idea to do an RTS set on a globe using polar coordinates (latitude and longitude), where you could spin the globe and zoom in and out to see the action. This was going to be called _MARS_ and was to be set millions of years ago when Mars supported life, prior to hundreds of years of increasingly devastating war that ultimately eradicated all life and evidence that there have ever been cities there. This was to have been on the Acorn Archimedes 310 computer which made use of the new RISC ARM chip. This had a performance of 1 MIP and it had 1 MB of RAM. Health issues forced me to set this aside, and by the time I was ready to continue I had a 486 Tower PC and was looking to develop something for that. From what I was able to tell the industry standard was to use C++ and as I wanted my game to make use of overlapping windows, so you had a desktop where you managed an overload of incoming intel, this led me to buy a book on Microsoft Foundation Classes, which used Charles Simonyi's hungarian notation for identifiers which put a series of single lowercase letters ahead of a name so that all the pointers would get alphasorted together as p came first. As I looked at Bjarne Stroustrup's book on C++ it became apparent to me that it was atrociously badly designed and would actually silently compile typos as the syntax was too overloaded. PASCAL wasn't like this at all as Niklaus Wirth had made it syntactically unambiguous with lots of SYNTAX ERRORS you had to get past, but once you did that your code was only wrong in not mapping to your intention. It worked. It was a legal PASCAL program. Then I realised that C++ wasn't even properly backward compatible with C so that undermined the whole point of having it. As I learnt more about its poorly misunderstood version of Alan Kay's object concept and came to read Bertrand Meyer's Object-Oriented Software Construction where he properly explained how his language Eiffel was constructed for large scale software engineering I grasped some of the appeal, but was still not persuaded of its validity as an essential paradigm. Cloning seemed better than Inheritance taxonomies. HAS-A > IS-A This led me to a long period of research where I examined more than 1,700 programming languages, including some that were PhD theses. There were a lot of old ideas which had been effective in their niche implementations, but for market reasons had not caught on. A multiparadigm language which harmonised their inclusion was theoretically possible and after doing twenty versions of the syntax, made especially unambiguous in its expression so that the lexer/parser were much easier to write, I was ready to start implementation. Then my mother became terminally ill and I had to be her 24/7 carer and had no time for anything else. It has taken until this week for me to be back to a position of being able to resume. I'm finally looking at my old notes from seven years ago. So, if I don't count that hiatus I have been working on the design of the tools by which I hope to boost my productivity enough to make my game (which has grown more ambitious in the meantime) for 25 years. I anticipate it will take me a year to implement my language, a year to make the suite of middleware tools that I will make my game with, and then two years to make my game which will be an asynchronous MMORTSFPSRPG set in the Milky Way. Basically a 4X. I recently bought an XBOX SERIES S to try out _STARFIELD_ as I thought it might be quite similar to what I was working on, but it wasn't. It was hard to say quite why the game wasn't fun even when you ignore the initial bugs it had at launch (which I didn't experience at all personally in my time with it, but appreciate were a major turn off on PC). I also didn't find that Hello Games' _No Man's Sky_ was fun. Nor did I enjoy _Elite: Dangerous_ that much, although that was doing a better job of being a space ship simulator and I liked that it had the whole of the Milky Way galaxy in there. I haven't tried _Star Citizen_ so I can't comment on it. From a business perspective it seems that Chris Roberts is using the _Squadron 42_ campaign that he had originally secured Kickstarter funding for as a "carrot" to string along as many funders as possible as he builds out his persistent universe MMO funded by the sale of very expensive ships to fans who are usually referred to as "whales". In a sense they are holding up the completion of the campaign, but another factor is that the game was probably always designed to be resource hungry, much more so than anyone's PCs, and could only run well on current bleeding edge PCs, and would only be financially successful if it released when the average PC gamer had a much better minimum specification, such as an SSD and lots of RAM and a 4090. This could become the case in a couple of years time. It depends of whether we get into a global recession. Maybe it could release the campaign, which is apparently finished but needs to be polished later this year (2024), with the economy for the PTU and missions, etc. all being worked out for the MMO in 2025 during a series of more workable larger population betas. I think the game as a whole will continue to grow through 2026 and get aliens you can go face-to-face with in 2027, as they are at the moment only mere concept art. This means that _Star Citizen_ will beat my game to release, and I will have spent 30 years to reach beta. Jonathan Blow's third game might not be this _SOKOBAN!_ title we see him working on. That might just be something like _Invaders_ which is used to test the features of his Jai language. He talks of spending 20 years on his third game and releasing it in episodes. I have no idea what the scope will be or the genre he has in mind. It might not even be a puzzle game. If he is thinking of doing something with similar scope to my MMORTSFPSRPG as a kind of "everything game" then he really does need Jai to be a replacement for programming in C++ as he would never be able to finish it with C++ as it is such a mess. Blow has spoken about how he can't rely on things being the same for 20 years so need his own tools, he may even need his own IDE and Windows system and OS / device driver Software Abstraction Layer before he is finished. Maybe his work can merge with some of the better work of Linux Debian / HURD as I forsee Windows becoming toxic to consumers in the medium term. Blow is right to say Linux is not good now, but by the time he has gotten around to write his own GUI and IDE based on a graphics API based on his own generic SAL drawing on common features of Metal and DirectX and OpenGL etc. then being able to say draw a texture map to the double buffer and then swap at Vsync is basically all you need to put pixels on screen and then having a frustrum with light sources could be the more sophisticated set of tools by which to bring an image to a window and that could ASSUME your system had hardware ray tracing built in, given the time scales we are talking about as PS6 will likely do hardware ray tracing when it launches in 2030. Now, you could just make your own graphics SAL be based around hardware ray tracing of procedurally generated texture maps with diffs that were cached and potentially streamed textures with alpha channels (like video overlays), and that way make that be the only thing, perhaps go back to using non rational B-splines (NURBs) do stuff so that you don't have to test to ensure there was no chance to drop through a mesh and have figures not default to a T-pose, and there is probably a lot of decisions Blow as the designer of the graphics SAL could make which would not map 1:1 to every feature of Metal, DirectX, OpenGL, Vulkan, etc. as well as what the PS6 is capable of. It might be fun to have hardware ray traced GUI. More work to set up the SAL on the back end, but the default window open would include a frustrum and a light source, etc. Do it properly, do it once, forget about having to fret over the details.
It is in this very talk that he spoke about how he would use the first two of 20 years of dev time for Game 3 to build a new language that he'd always wanted for years so.. yeah, it shouldn't have surprised anybody thatactually watched this talk
The No Man's Sky reference at 44:50 hasn't dated well. Maybe the reason people shy away from ambitious vision these days is because it's extremely hard to deliver on and when you miss, which most people do, you get demonised?
No Man's Sky flopped because of the content, not because of the vision. They had solved the hard problems like generating an entire universe and coping with the geometry of "real" planets. The Design however was one of the worst. This is not an example of how achieving ambitious visions is extremly hard, but one of operational blindness. I mean, have they played their game? whatever. The fact that no mans sky was shit does not lessen the strength of the argument.
I actually thought the game was fantastic. I've dreamt of playing a game like that since I was a kid. A game where I was able to jump in a space ship, fly out of the atmosphere, see another planet, fly and land on that planet and get out of my ship and explore. The atmosphere was lonely and I just found it to be perfect.
Also, I don't know how true it actually is... but there were rumors that they ran out of funding and Sony wanted it pushed out. So it could of been that they just simply couldn't get it to the stage they wanted it to be. And if this is true then it's another reason that strengthens Jon's reasoning on why his project had to be self funded and owned.
![[2019 TGDF] Making Games in 2019 and Beyond (Jonathan Blow)](/img/n.gif)
I could listen to this guy talk all day.
2:26 Start
awesome video!
Love Blow, but where is that "Game 3" he is talking about almost being finished if this was from 2014?
very inspiring talk
jblow: the roguelike
Watching this 6 years later, I can't help but think of Star Citizen. It certainly won't be on the same level as Gravity's Rainbow, but I think that the fact that there is constant backlash around its development time demonstrates that most gamers do not even support the idea of an ambitious 20+ year project, and certainly no publisher does either. Maybe the saddest thing about the 21st century as a whole is that we've lost the ambition to do big things... it may not be in everyone's department to work on 20yr+ projects, but damn, that department should be the one with the most funding and respect instead of the least funded and the least respected.
I had an idea to do an RTS set on a globe using polar coordinates (latitude and longitude), where you could spin the globe and zoom in and out to see the action. This was going to be called _MARS_ and was to be set millions of years ago when Mars supported life, prior to hundreds of years of increasingly devastating war that ultimately eradicated all life and evidence that there have ever been cities there. This was to have been on the Acorn Archimedes 310 computer which made use of the new RISC ARM chip. This had a performance of 1 MIP and it had 1 MB of RAM. Health issues forced me to set this aside, and by the time I was ready to continue I had a 486 Tower PC and was looking to develop something for that. From what I was able to tell the industry standard was to use C++ and as I wanted my game to make use of overlapping windows, so you had a desktop where you managed an overload of incoming intel, this led me to buy a book on Microsoft Foundation Classes, which used Charles Simonyi's hungarian notation for identifiers which put a series of single lowercase letters ahead of a name so that all the pointers would get alphasorted together as p came first.
As I looked at Bjarne Stroustrup's book on C++ it became apparent to me that it was atrociously badly designed and would actually silently compile typos as the syntax was too overloaded. PASCAL wasn't like this at all as Niklaus Wirth had made it syntactically unambiguous with lots of SYNTAX ERRORS you had to get past, but once you did that your code was only wrong in not mapping to your intention. It worked. It was a legal PASCAL program. Then I realised that C++ wasn't even properly backward compatible with C so that undermined the whole point of having it. As I learnt more about its poorly misunderstood version of Alan Kay's object concept and came to read Bertrand Meyer's Object-Oriented Software Construction where he properly explained how his language Eiffel was constructed for large scale software engineering I grasped some of the appeal, but was still not persuaded of its validity as an essential paradigm. Cloning seemed better than Inheritance taxonomies.
HAS-A > IS-A
This led me to a long period of research where I examined more than 1,700 programming languages, including some that were PhD theses. There were a lot of old ideas which had been effective in their niche implementations, but for market reasons had not caught on. A multiparadigm language which harmonised their inclusion was theoretically possible and after doing twenty versions of the syntax, made especially unambiguous in its expression so that the lexer/parser were much easier to write, I was ready to start implementation.
Then my mother became terminally ill and I had to be her 24/7 carer and had no time for anything else. It has taken until this week for me to be back to a position of being able to resume. I'm finally looking at my old notes from seven years ago. So, if I don't count that hiatus I have been working on the design of the tools by which I hope to boost my productivity enough to make my game (which has grown more ambitious in the meantime) for 25 years. I anticipate it will take me a year to implement my language, a year to make the suite of middleware tools that I will make my game with, and then two years to make my game which will be an asynchronous MMORTSFPSRPG set in the Milky Way. Basically a 4X.
I recently bought an XBOX SERIES S to try out _STARFIELD_ as I thought it might be quite similar to what I was working on, but it wasn't. It was hard to say quite why the game wasn't fun even when you ignore the initial bugs it had at launch (which I didn't experience at all personally in my time with it, but appreciate were a major turn off on PC). I also didn't find that Hello Games' _No Man's Sky_ was fun. Nor did I enjoy _Elite: Dangerous_ that much, although that was doing a better job of being a space ship simulator and I liked that it had the whole of the Milky Way galaxy in there. I haven't tried _Star Citizen_ so I can't comment on it. From a business perspective it seems that Chris Roberts is using the _Squadron 42_ campaign that he had originally secured Kickstarter funding for as a "carrot" to string along as many funders as possible as he builds out his persistent universe MMO funded by the sale of very expensive ships to fans who are usually referred to as "whales". In a sense they are holding up the completion of the campaign, but another factor is that the game was probably always designed to be resource hungry, much more so than anyone's PCs, and could only run well on current bleeding edge PCs, and would only be financially successful if it released when the average PC gamer had a much better minimum specification, such as an SSD and lots of RAM and a 4090. This could become the case in a couple of years time. It depends of whether we get into a global recession. Maybe it could release the campaign, which is apparently finished but needs to be polished later this year (2024), with the economy for the PTU and missions, etc. all being worked out for the MMO in 2025 during a series of more workable larger population betas. I think the game as a whole will continue to grow through 2026 and get aliens you can go face-to-face with in 2027, as they are at the moment only mere concept art. This means that _Star Citizen_ will beat my game to release, and I will have spent 30 years to reach beta.
Jonathan Blow's third game might not be this _SOKOBAN!_ title we see him working on. That might just be something like _Invaders_ which is used to test the features of his Jai language. He talks of spending 20 years on his third game and releasing it in episodes. I have no idea what the scope will be or the genre he has in mind. It might not even be a puzzle game. If he is thinking of doing something with similar scope to my MMORTSFPSRPG as a kind of "everything game" then he really does need Jai to be a replacement for programming in C++ as he would never be able to finish it with C++ as it is such a mess. Blow has spoken about how he can't rely on things being the same for 20 years so need his own tools, he may even need his own IDE and Windows system and OS / device driver Software Abstraction Layer before he is finished. Maybe his work can merge with some of the better work of Linux Debian / HURD as I forsee Windows becoming toxic to consumers in the medium term. Blow is right to say Linux is not good now, but by the time he has gotten around to write his own GUI and IDE based on a graphics API based on his own generic SAL drawing on common features of Metal and DirectX and OpenGL etc. then being able to say draw a texture map to the double buffer and then swap at Vsync is basically all you need to put pixels on screen and then having a frustrum with light sources could be the more sophisticated set of tools by which to bring an image to a window and that could ASSUME your system had hardware ray tracing built in, given the time scales we are talking about as PS6 will likely do hardware ray tracing when it launches in 2030. Now, you could just make your own graphics SAL be based around hardware ray tracing of procedurally generated texture maps with diffs that were cached and potentially streamed textures with alpha channels (like video overlays), and that way make that be the only thing, perhaps go back to using non rational B-splines (NURBs) do stuff so that you don't have to test to ensure there was no chance to drop through a mesh and have figures not default to a T-pose, and there is probably a lot of decisions Blow as the designer of the graphics SAL could make which would not map 1:1 to every feature of Metal, DirectX, OpenGL, Vulkan, etc. as well as what the PS6 is capable of. It might be fun to have hardware ray traced GUI. More work to set up the SAL on the back end, but the default window open would include a frustrum and a light source, etc. Do it properly, do it once, forget about having to fret over the details.
Competency. Just have 1 billion people on Earth we will reach Kardashev level III in 1 century
Designer - KOJIMA-san
Well, it's mid-2018, where that Game 3 at?
Have you seen how much work has gone into Jai?
Wait, he is developing a programming language now? Yeah, no, why am I not surprise? Oh, well, back to ignoring him again for a few years.
It is in this very talk that he spoke about how he would use the first two of 20 years of dev time for Game 3 to build a new language that he'd always wanted for years so.. yeah, it shouldn't have surprised anybody thatactually watched this talk
this guy is the 'brute lee; of game design
We call him Bruce B!
so you want to be toady? excellent choice
Right? Btw, if Tarn isn't the Bruce Lee of game design, he is probably the Bruce Lee of procedural world generators design
@@forcefield6973 I mean no disrespect but if he were the Bruce Lee of game design, he would be done with that game by now.
The No Man's Sky reference at 44:50 hasn't dated well. Maybe the reason people shy away from ambitious vision these days is because it's extremely hard to deliver on and when you miss, which most people do, you get demonised?
Perhaps so. I've checked Project Hieroglyph's website too and there doesn't seem to be a lot of updating on there these days.
No Man's Sky flopped because of the content, not because of the vision. They had solved the hard problems like generating an entire universe and coping with the geometry of "real" planets. The Design however was one of the worst. This is not an example of how achieving ambitious visions is extremly hard, but one of operational blindness. I mean, have they played their game? whatever. The fact that no mans sky was shit does not lessen the strength of the argument.
You're right, ambitious games are a bad idea, and always suck. We should only try to make unambitious games then.
I actually thought the game was fantastic. I've dreamt of playing a game like that since I was a kid. A game where I was able to jump in a space ship, fly out of the atmosphere, see another planet, fly and land on that planet and get out of my ship and explore. The atmosphere was lonely and I just found it to be perfect.
Also, I don't know how true it actually is... but there were rumors that they ran out of funding and Sony wanted it pushed out. So it could of been that they just simply couldn't get it to the stage they wanted it to be.
And if this is true then it's another reason that strengthens Jon's reasoning on why his project had to be self funded and owned.
"Right?"
nice high level observation - glad you were paying attention buddy