Classic Game Postmortem - Marble Madness

[Music] well thank you all for coming to one of the GDC 25th anniversary post-mortem we're here today of course to talk about Marble Madness it's not going to be a whole our talk I was told to leave 20 minutes for questions so I'm targeting 40 minutes with it one request from the GDC is please fill out the evaluations they're very important to the overall process thank you tough crowd so I've been making games for about 30 years now and way back at the beginning of my career I made three games worked on three games I should say during sort of the final years of the Golden Age of the arcade and the third of those games was Marble Madness now I hadn't really played arcade games much at all to begin with in fact pong I don't know it seemed kind of limited and

plus you could play it at home but a friend of mine came to me in high school and he said mark you have to try out this game it's the coolest thing you're you're shooting at soldiers who are advancing at you and shooting back and you're you're dodging behind bunkers and things he's actually talking about space invaders I don't quite see the soldiers there but anyway I did play it and it was indeed the coolest thing ever and so I played starting to play a lot of arcade games and I loved more than anything else the variety of these games because these are games from 1980 and 1981 eight games but also eight completely different genres that's the way it was in the arcades those days if you'd go in every week and you'd find something totally new you never seen before now in those days I was also a pretty big paper RPG fan I used to play them with my brother and his friends ngo paper RPGs you draw a map and then you have to make a copy of part of that

map and show it to somebody to show them where they are and you have to roll a lot of dice and my brother and I thought that hey wouldn't this be simpler on a computer and so we found a couple hundred thousand dollars of equipment to the nearby university that was underutilized and we sat down to make a 3d real-time action RPG in 1977 and the idea was that as you were wandering around the dungeon above-ground below ground that the graphics would be like battlezone but better because they'd be they'd have a solidity to them we were actually going to do we're trying to do hidden line vector which is a very tricky kind of graphics and we could do about on our scope wasn't even a monitor on our scope about one image every few seconds and then for for the action sequences we were going to shift to sort of a cinema Tronics Warrior style

overhead real-time combat simulator now my brother and I had about as much chance of finishing this project as to six year old kids in the backyard trying to dig a hole to China but I did learn a lot about programming and oh and we did see the game we were trying to make by the way it was called Final Fantasy 7 many years later so I had two hobbies one was programming and yes it was on punch cards and the other one was playing arcade games and I used to dream about turning my two hobbies into a job and getting fame and fortune Fame because I could see it I'd release that game in the arcades and then I'd be on Johnny Carson talking to him about you know my latest creation and as for the fortune I figured wow that quarter if I could just get five cents out of that quarter I could just get one penny out of that quarter you could do the math that would be like millions of dollars and I got very lucky is while I was in

college I ran into a man who was writing a book on arcade strategy we're talking an era where you got your information from TV radio newspaper or magazine that was it we're talking pre-web we're talking before the release of the CD here so if you are going to get information on an arcade game and how to play better you were going to read a book and so I ran into this guy and he interviewed me because it was a very good arcade player and after that he offered to set me up with an interview with the head of product development at Atari and January 1982 I became employee 16-thousand of Atari more properly speaking I was in a much smaller part of Atari Atari with three different companies there was a company making home computers a company making consumer game machines and consumer games and then a very small company making arcade games only about 150 people in this group and that's what I joined at the age of 17 years and four

months I quickly learned that Wow oh here's me at the time by the way I quickly learned it well and the thing about Fame it wasn't going to work that way we were forbidden to tell people that we had created these games the stance was that it was a corporate creation so my good friend Dave toyour released Tempest it sold 30,000 units all across North America and he couldn't tell a soul that he created that game basically by himself now the consolation you got for this was that you could put your initials in the high score table and so you can see the creators of centipede here Don Bailey and Ed Logue and their initials there that's the same and therefore the fortune well turns out the economics of the arcade business weren't that great so we made the game but we didn't get those quarters what we did was we dad a couple hundred dollars in margin and sell it to a distributor who would sell it to an operator who would perhaps have

a arcade like time zone and that operator would get a hundred percent of those quarters but would have to pay for rent and salaries and maintenance which was huge actually those joysticks would break down all the time alternatively the operator could put the game in a 7-eleven in which case no salaries no rent but only half the quarters and either way the economics were not great the operator needed 10,000 plays at 25 cents to purchase a $2,500 machine except that math assumes there's no overheads at all realistically speaking the operator needed 20,000 or 30,000 plays to break even so nobody was making very much money now the final thing was I kind of knew this intellectually but I didn't think it through with I wanted to make 10 or 20 hours cinematic story based games and I was in a world right one quarter got you three minutes it doesn't get you this whatever it is and I guess the only

consolation I had was that we were totally empowered at Atari there was one job it was the programmer / designer / artist so you you made the design doc out of the prove did the artwork put it in the game coated it up even as a rookie you do this and consequently there was a hundred percent failure rate for first titles even a famous guy like Dave toyour his first title was a soccer game that was unsaleable low in income when he put it in the arcades trivia true trivia here it was a soccer game based on a track ball and the track ball was actually a bowling ball and this by the way this failure rate finally broken in 1983 by France LAN Xang or who I hope is here today are you here from stand up this man the first man whose rookie title was published and it was of course

crystal castle so 20 or 30 before couldn't do it France could do it there was also something going on here in that you control your own destiny you had a very direct link to the consumers these titles were not marketed at all they couldn't be marketed all you did was put them in the arcade and see how much money that they earned and the players you know they could see it for free they could play it for a quarter if it was fun they'd play it again if it wasn't fun that was it or maybe they played it ten times and the tenth time wasn't fun and that was it it was very very tough business talk about monetization hell and to see if a game could be published what we would do is we makes a game about up to the 80% point so it looked indistinguishable from a finished game and we put one of them in an Arcadian and to see what the coin drop was and at this point two out of every three games would be canceled it's just it was a

brutal environment to be in so here's my first game which is one of those two of out of the three that did not get published thank you you can play it on main it was a what was it you'd slide tiles and try to get the ducks to safety on the right-hand screen or the bottom of the screen and true form I did every piece of artwork in this game except for the duck and believe it or not this thing was viewed as great artwork for its day but very poor gameplay so after this failure I sat down in early 1983 now that I had a little bit of experience in the games business and tried to figure out what sort of game I wanted to make and there were a few design targets that you had you needed to make something unique sequels did not sell believes or not Atari was particularly serious about making unique titles if there was ever a

top-down driving game that had been released in the history of games you couldn't make one if there was a head-on head fighting game that had been released you couldn't make one because that would be insufficiently original so the very high bar to hit with these games the other target was unique control because centipede and asteroids and tempest of course all of these had control we believe it was an important part of keeping the concept fresh the final thing which started coming out in 1983 was that the game should have at least two players because the economics were getting worse and worse and two players at once meant two coins so I thought about this I love 3d graphics and there was the battlezone and that kind of hardware that could do it but it was still floating I mean these objects are transparent and there was this other hardware in development at the time since actually before I came to Atari called iRobot and iRobot was raster 3d

so it's solid these images in 3d now very low polygon count about 5,000 polygons a second and so the poor designer here was trying to try to make an entire frame of the game with 200 polygons and the other thing about this is when you really looked at it all of the edges of the polygons would crawl you get very distracted by the aliasing in the game and I wanted to be the first person to do solid 3d so maybe not a lot of moving camera but really clean 3d and I sat down to figure out what that game would be and how I would go about making it and actually front beat me to it with Crystal Castles but we won't talk about that today I would sit around late at night on our pixel processing system so this is our art system we had two of

them in the office it was built on a mini computer about the size of a refrigerator and I'd sit there and I'd fiddle and I thought okay steam how about miniature golf people like that and then what do you do in the game well what are the controls before that sorry I had to work out sort of how I was going to do the contours of the play field I worked out that two lines in different directions could give me the contours and I hold I don't know if that would work so I put pipes in there and this is a actually a copy of our old design document I could not take anything with me when I left Atari they were not really all that Creator friendly and when Atari finally closed in 2004 a man named Scott Evans went there that day as they threw out everything from the 30 years of company's history picked it up and all of the stuff is on his website Atari games com I'm very grateful he did this because the only copy of my own design document

I have is from the website so for controls we'd actually done for quack one of the versions we done a industrial-strength touch screen and the idea was this was a frame embedded in the cabinet shooting rays of light horizontally and vertically across the face of the monitor so if you put your fat finger through to the face of the monitor it could detect where it was and I thought okay well we'll use that touch screen and you'll somehow hit the ball and you got to go around corners maybe you'll place bumpers or something to help yourself around the corners and I don't know what is the second player do not really clear I walked this around the office and everyone hated it so then I thought what about erase people like to do that and what about we use a trackball list of control so you'd see a ball on the screen and have a ball to roll but what about a motorized

trackball where we'd actually have the ability to spin the ball programmatically so for example as if the ball went down a slope faster and faster you could just stand there and watch that ball roll and if you wanted to break it to put your hand on it if you put your hand on it and it's on a slope you can tell what the slope is just by the way it's fighting your grip and then for actual gameplay a race if you get to the finish first you continue if you don't game over and for two-player play I thought how about two players versus two AI opponents and I thought how about the players are glass marbles and the AI opponents are steel marbles or steel this plan was proof but first you had to make the hardware because every game Hardware in those days was unique to the

game for which it was created so clearly you needed to do that for something like Star Wars versus from something like centipede but the truth is that no matter how similar the Atari games look to you as a player actually with all custom work inside the cabinet so it takes a while to make hardware and while they were doing that I went off and helped my friend Owen Rubin on his game major havoc and the hardware engineer started on the marble madness hardware now I had to specify that as well and I had again two choices vector or raster pretty much everything those days was one or the other vector was simply colored lines on a black background but the plus was they were under complete programmatic control so you had a command buffer like a modern graphics hardware and the program would just put the list of lines in it anything you wanted to draw you can draw and if you wanted 3d they could put it what they call the math box on it and all of a sudden the images would look like that

with raster Hardware so they were based on wrong and they were incredibly finicky you were not able to do any freeform graphics whatsoever you had a ROM back up there you had a background that was made out of 8x8 pixel sense that was a basic unit that you populated with and then you've had a very few sprites maybe 16 or 32 which were small movable objects on top of that that also came out of the wrong and every single graphic you were going to display had to be in a ROM in custom chip basically and so these are stamps from centipede the images you could display in centipede that was it times about ten and what you did was you pass with the hardware the code for that stamp you did not have the ability to say I want to put a dot on the screen and put it here now during development we couldn't afford the six week turnaround time to

make custom ROM so we used erasable program realms EP realms just an example let's centipedes background it's all made from two types of stamps if you want the mushroom to disappear you have to just substitute things that look smaller and smaller until it's gone and then something like the spider is one of those sprites on tops that you have maybe 16 or 32 of and the only thing you can do run time programmatically is change the color except everything else is can't so this created a situation where a blog if he wanted to put a asteroid that looked like a mushroom in his game on a vector hardware that was a couple minutes type it in download it done whereas if Ed log thought that the legs on the centipede should be one pixel longer he had to go lengthen them on the pixel processing system transfer the data to the main mini computer that we used running utility put the results on an 8-inch floppy take that to the

EEPROM burning room enter a code burn a set of roms go back to the lab lever them out put in the new ROM and then find out that probably he screwed up one step along the way so ours very inflexible so there was a third choice it was a very unique choice the irobot hardware if the very clever hardware engineer liked my game design and he said you know mark these procedural play fields that you want to make are really easy to do in hardware why don't we make a cheap version of the irobot hardware you'll just give it the heights you want to display and it'll do it live and at that point I got really excited because now the play field can be sort of a living creature that's a part of the gameplay can chase the player around and that was so cool I put this in the game design so the core game design concept then became and a race against an AI

opponent on this dynamic play field and what happened next is that of course this didn't happen the economics of the arcade business got still worse we were having a crash in 82 and 83 and 84 we lost 90% of our sales it's very tough weed saturated the market there were one and a half million arcade games in North America in 1982 and we just couldn't sell anymore the old ones weren't dead yet so our strategy for for improving the economics of the arcade business was we were going to do system hardware system Hardware meant that the operator would buy a 25-year hundred dollar game machine one time and it would have a fit what we call the system board in it and after that the operator would buy new games just by getting a giant cartridge like I have a photo of here doesn't really look like a cartridge right now and a control panel

specific to that game and Plexiglas giving the game's title and maybe some artwork to the side of the cabinet and we figured that that would be about eight hundred dollars rather than twenty five hundred so we would help out the business by not having custom hardware every time now we were in luck because there been a skunkworks project again going on from before I joined Atari where a poor engineer locked in a room was embedding all of the conventional stamp graphics in a single custom chip and so he'd come in to work every day and manipulate diagrams that look like this it's apparently how you made chips in those days it certainly isn't how you do it now and so we would have our ROM stamps and our ROM sprites and in addition to that we would have RAM sprites and that meant that I would have a programmable area maybe not large 32 by 32 pixels 64 by 64 pixel and

inside of that I could draw any graphic I wanted to programmable and that meant marbles that could refract the play field behind them and feeley's that could reflect the play field around them and this totally did not happen as well it did not happen because we blocked the guy in the room too long and he couldn't take it he quit he took another job he told us it was one month just one month to finish the chip it'll be fine well we looked at it it wasn't a month it was six months through a year to finish that thing and it wasn't just me that was waiting for hardware by this point we as a whole company we're going to go to system Hardware so all the teams were waiting so very reluctantly the engineer settled down and they did what they call the budget system the budget system was a UH turley conventional ROM based hardware and you can see here the motherboard on left that's the marble madness cartridge on the right only about two things notable

about this Hardware fifty-six sprites so a few more and if you wanted 32 or 64 colors rather than 16 you could have that many you just needed to put more roms into the cartridge the motorized trackball had also been in development now before I get into a motorized trackball let me show you the inside of a basic trackball it's supported on three shaft so there is a shaft with an optical sensor to detect the vertical motion there's another one to detect the horizontal motion there's a third shaft just with the trackball which is actually a cue ball from billiard can make good contact with the mechanisms so these guys have made a prototype with with four shafts they had two for detection and two for driving the ball with motors and I found out there with four shafts the cue ball couldn't make good contact with all of them simultaneously and for anybody who's

ever bought a kitchen table dining room table you know this right if it sits on three legs it's stable if it sits on four you need shims or something and they could not machine this to enough precision that it would survive in the arcades and so this too was scrapped right here is the actual marble madness control panel and it is too utterly conventional drywall so 1984 the hardware finally arrives and let me tell you a little bit about our development environment here we were using a backs mini computer the 11 780 it was a true one nip machine that means 1 million instructions per second this does I think 10 billion I'm not sure not a whole lot of horsepower that might have been okay but the thing is 50 of us shared this machine on terminals so whatever you wanted to do you were going to do very slowly on the good side Wow the 68000 CPU I got to be the first

person at Atari to use it and all I can say is it was posh 32-bit registers eight of them eight address registers and it ran at an amazing seven megahertz it was it was great we were also the first team to use C as a programming language and I was a little little excited and a little suspicious you know what's going to happen if I don't have direct control over the assembly language instructions and that was actually a justified suspicion is sure during the beginning part of the project you could type something in it would compile quickly and you could see the result but after a while it was taking 15 minutes to get through the compiler linker and download process to see even a single character change in the game and there's also this hidden time think in that I had to go back in and hand optimize a lot of the code the C compiler created net now we we ran at 30

Hertz because we use C not 60 and I suspect that it actually took longer to make the game because of that now I've been saying we and I just said that the you know this was the error of the one person one project well by 1984 that had changed it was a three-person project the second of those people was Bob Flanagan who had been helping out John Saul what's on paper boy programming and then came on to be my partner programming on Marble Madness and so we settled down to code the game in C together in our first milestone remember the pitch player versus AI opponent on a dynamically changing terrain and so we had to somehow show that we could use a rom hardware very inflexible ROM hardware to do this kind of smooth graphics you know iRobot no problem but

if it has to be a canned graphic in a stamp big problem so I looked at the playfield and this is uh this is bump and made out of stamps there's a lot of duplication here all of the flat areas are just those two stamps right there there's all the flat areas and then those stamps are actually the same stamp and if we look at the front of the bump it's the same stamps repeated there's not that many unique stamps needed to do a bump like this but what stamps do you need is next the question and so just looking at one stamp here that stamp I realize is trying to represent three planes of the bump behind it so zooming in on it there's the three planes you can represent really the stamp as that zigzag line and you can give it a number two you can say it starts on the left and it starts ten pixels up above the the stamp and goes to the right six

pixels up comes down it's four pixels and so on you can turn any stamp into a couple numbers so here's another stamp and so I thought okay I will sit down I'll figure out what the static aspects of the playfield will be I'll derive the library of stamps I needed to represent that I'll take four hours to burn that into roms I'll put them in the game I'll play the game and it won't have all the stamps it needs for that living bump as it chases the player around it'll use the closest one it can find but it'll keep track of the one it doesn't have then off line I'll incorporate those additional stamps into my library and reburn the wrong and after a couple cycles the idea is everything you need for a dynamic play field will be in that row and we did on this basis a simple proof of concept and it appeared to work now the second milestone was the player versus the AI opponent race

and Wow yeah the design document has all been about the competitive and the cooperative aspects of racing competitive is things like the first bowl breaks the glass supports for the trap and the second ball is trapped by it or there's a cliff that you can't jump down your marble would shatter and so the first ball can use the teeter-totter to get down but the second ball has to wait and so on and here is the actual play field from the design document you can see it's a symmetrical track for the player and the AI opponent until you get to a wedge sitting on glass supports and the first person to hit that those glass supports breaks them takes the shortcut and forces the other player to take the long way around so that was competitive then for cooperative the idea was you and I play against two computer opponents but maybe

there's an elastic barricade where you and I have to hit it simultaneously if we do we get through but if I get there before you do I just have to wait for you to get there so we got the race in against the AI opponents we got the dynamic play field and we put it all together we played it and it was a enormous disappointment the bumps that chased you around it just felt creepy it had a weird rat under the rug kind of feel to it and it didn't really affect the gameplay very much and as far as racing against an AI opponent went the thing was you'd be going as hard as you could but we couldn't let you go too fast I mean these play fields couldn't be very long and minut variations in how you went down trenches and around corners would be all that determined who won the race it was a pretty complete disaster now luckily the my second

partner on this Sam Comstock our artist we had a full-time artist it was great he'd been doing the critters for the game and he looked at this and he said no no static is actually good if you you get rid of the dynamic concept and you the playfield static you can do shadows which will make it pump and you can do ornamentation which will make it prettier and a programmer friend of mine said and if you retrace it you can anti-alias it which will get rid of the Jaggi on the edges of everything so to start with the last of those things the concept here was I would define a height field I would then write a program that would Ray trace it on our shared mini computer 16 rays for every pixel that we would then average and then I would hand it off to the artist to touch up by hand and as far as ray tracing I don't mean anything particularly complex this is a play field seen from above I meant take a ray in from the front see if it hits a

white center of a square or a black line and figure out the color there then go from the point of collision out to the light source if you can get to it you're in sunlight if you can't you're in shadow and as far as more rain 16 Ray's is definitely better than 1 ray you know you got one ray it's going to be light or dark and you'll see like in this ramp area some significant jaggies if you can have a lot of rays and average them we use the term it'll filter for that you can get these very I don't know delicate looking graphics quite unique in the day so that worked very well as far as shadows went I went in and I doctored last week one of our screenshots to remove the shadows to show you what they look like without shadows so the left is marble madness as if it had no shadow and it actually it's quite hard I'll zoom in it's quite hard to see the

relationship spatially between the objects here if you look at the pillars on the right it's a little hard to tell what the height of each of those is but as soon as we add shadows you instantly know the spatial relationships there so yeah Sam Comstock was correct and then of course ornamentation sure you get your artist in there and he freehand I don't know rails and their shadows and arrows and adds color and of course it's going to look better the problem we ran into was I needed to trace eight eight seven excuse me just seven seven million raise a level on our shared computer that was twelve hours of full-time usage on a computer shared by fifty people so I did nights I did at weekends France has forgotten I mentioned it yesterday to him he bursts into my office some Saturday morning why do you know what the hell I was doing because what we

didn't know is the sharing algorithm on the computer if you use the whole mini computer for too long you don't and nobody else could get in it it took him 15 minutes to type in his name and password to log in at any rate I did senesce that problem I did become the largest user nights and weekends of this computer and we did by this whole transition from dynamic - static we did lose a month but we gained a look as far as the race versus AI opponents went that turned into a race versus time I just loved pole position to death so I simply used the pole position time algorithm it's very simple you race if you have seconds left over at the finish line you keep them and you get more seconds so if we look at the transition between early 83 when I did the paper design doc and June of 84 when we actually got in the game up and found its problems it was a completely different game however that was a plus

we had this killer look and we also had a pretty good game play here with race versus time that turned out to work very nicely and intuitively and finally I haven't talked about this much but we had this really nice approach with the enemies that I understand had worked out where there were all bits and pieces of things like this link these are the marble muncher is just a foot or the O's is just a blob no eyes or anything or the birds are just things with wings I should also say we had a great audio team who they come by occasionally and download stuff on it and it was great stuff invariably I had nothing to do with that creative Brad now they they got the Yamaha FM sound synthesis chip it was from a color organ this was I believe the first time that ship had ever been used in an arcade game it had just beautiful audio and they did their excellent sound effects and sort of sort of pseudo classical compositions so it was

actually from that point I know this is boring but it's pretty smooth sailing less than six months from that time to release we made three levels and we did a in-house focus test where we just had players come in and play they'd loved it we made two more levels and we put a game machine a prototype in an arcade near my house and a lot of a lot of coins drop a lot of income we added one more level and we shipped it just for brief stories from game development I have the original design sort of a black hole style gravitational funnel where the force would suck you down into it and Wow I couldn't get that to work so and it was very hard to iterate on the runs it's not like I could try any more than one shot at this a day because I

had to rerun the levels overnight and after four or five tries I just gave up and the dimple you see in the second level is what we ended up with I ran out of sprite so we had marbles the marbles had dots in them to indicate their motion and I needed eleven sprites to make one marble because I needed the left half the right half eight dots just to do a dot you needed to use a sprite and then I needed a highlight that would be in front of the dots two marbles twenty-two sprites we had 56 total thought about this I ended up making every single possible combination of two dollars and by using that library I could then do it with four sprites that each had two dollars and save eight sprites and that got us through our right crisis the ooze we really wanted to do a translucent effect and we did

not have hardware that supported translucency so Sam drew the grid into the booze it's very well done we had a bit of a kerfuffle with our graphics designers this was the age of the the character pac-man right biggest game of the day pac-man I didn't want to do that I wanted to do this abstract SRS game where nothing had eyes and they looked at the marble listen how that's not a marble that's a happy face and they did the Plexiglas top with happy face marbles and then did the control panel with happy face marbles for our first tab and I negotiated this one as you can see we ended up with a smile but not a face so right those are those are happy what are they yeah they're happy marbles they're not happy faces so we shipped it in December of 1984 we

sold 4000 units the market had come way down in the heyday was a hundred thousand for something like pac-man even something that was a failure perceived as a failure like battlezone sort of sold 30,000 1982 pole position sold 20,000 1983 millipedes sold 10,000 by the time this game is time came out that was one of the top sellers of the year at 4,000 units what was good about that was we were in such trouble that kept our factory open for three months otherwise we would have had to furlough the workers we were number one in income in the arcades we tracked across the board we tracked 18 arcades and after six weeks simultaneously dropped out of number one because the game was four minutes long yeah I've come back why so short it really doesn't make a whole lot of sense and Scott Evans Atari games comm sent me my source code I did not ask him for it

he sent it to me I looked at it I should not have I felt so small the cut for the ooze is 300 lines of C I mean we're talking not even student project level for this and we were professional programmers working at the top game company in America and arcade games I don't know it's just mind-boggling I looked at uncharted last week the AI code for typical enemy in uncharted is 750 lines that code is written by a game designer not programmers so all I can figure is that our environment had so many obstacles it threw in front of us you know four hours to see your graphics on the screen and those compiles and the shared mini computer and the like that it just to get a 64 K program was an accomplishment anyway I'm going with that it wasn't all bad that Marble Madness had such short

income I mean it was a system Hardware and we were the first game so as soon as that dropped out there were more games ready to sell to the operators like Peter pack rad and Roadrunner and Indiana Jones I like one final thought Marble Madness is certainly the most creative title I've ever worked on it's probably the most creative title I ever will work on there's a certain commitment to a concept there that hasn't been anything else I've ever been on and that that's really rare if you if you look at the games business today there's not so much of that couple exceptions aside and in the process of putting this together I was thinking about it what I think is this is our games were so small in scope that you got one concept and one concept with asteroids and spaceship go or mushroom and centipede go and it would either work or it would not work but simply we were in an environment that fostered

that required that kind of commitment to a creative idea and that's really it for the prepared talk thanks so much and I've been asked if you have a question to use a stand like hi great talk if you were to iterate on Marble Madness today and take it to the next level what would you do I want to be sort of a philosophical quest where the marble goes out into the real world and has conversations with people about things like why's there evil in the universe I'm really not kidding I had you to design for this a couple of years back it will never happen oh yeah this one goes back a little while so understand if you don't have the answer for this I read in the press 20 plus

years ago that you were in you are unexpectedly a detriment to the major havoc project because you were such a good play tester the game was tuned to high in the arcades I read that in an article and I don't know if that was someone being they were talking about you in the context of Marble Madness yay yes at the arcade show last year the Gravatar team said I almost took the team down actually because yeah I would be in house play tester and I thought it was much too easy it's a good less than usability for everybody involved let me tell you that I haven't heard that about major havoc so we we just have to kill the players to jarn's hash with the thing I mean if the game time play averages weren't three minutes or less it was unsaleable and major havoc because it was sort of a platformer in some respects was more like a four-minute game a question about all the follow-on ports of Marvel minutes have already would involved in any of those bringing or - a and the platforms and did it have any kind of a different

what it was a contrast really notable - you know I wasn't involved in a those it's really a shame that they couldn't have used trackball it's not really such the same game as a joystick but now that we have tablets and things we might be able to use the gyros in there to do something well I could have somebody show me their version of marble madness on a smartphone one two and asked if I wanted to convert it to other smartphones and I noticed they dripped my credit out of the game and my question was if I did the project would I be responsible for ripping out my credit or with somebody else take care of that for me I remember when I played a game in the arcade I was always very convinced by how solid the physics felt I was just wondering if you had any comments on how you implemented the physics for the game oh wow with a hack thank you for the kind words though yeah you kind of roll the ball and he got to an edge the edge would push you off usually I'm sure you've seen the speedruns where you take advantage of the bad physics to roll off

and then pitch into an edge and just pop up to the play field you can do that almost anywhere in the game putter's so at any point during the design process did anybody mention that the gap between the trackball and the control panel is just perfect enough for a seven-year-old to get his uh his middle finger stuck right in there now you should have played dave tires game where it was a real bowling ball without holes because you could get the entire side of your hands in there you owe me some band-aid money any more question no we might yeah I can repeat it yeah so the follow-up game Marvel man that was Bob Flanagan with the other programmer on marble badness working with oh he'll kill me because

I'm forgetting his name hailey who worked on paper boy and a couple other very famous game and they my khalasar and they took a very different approach they thought that the sort of abstract look was hurting them so they went with enemies like a tomato or giant fork and knife and crack levels were actually terribly expensive trackballs worth $50 a pop and they break down frequently so they went with joysticks so they could afford to put three of them on the control panel I took it all the way to test and found that the income was not high enough to release yeah I was just wondering when you after you rendered the playfield and you broke them into up into tiles did you ever run into the case where you had too many and how did you resolve that finish I came to die too many unique tiles so now I kind of just worked and I didn't think too hard about it I wish I had a better answer than that I

should have done more play fields than that I was really a mystery looking back why that is all that is in the game when we could have run you know three months long longer and had an eight minute game and maybe not dropped out of first place in the arcades after six weeks thank you going to ask if you are sort of a very untechnical question if you have any favorite game from the era back then and a favorite game sort of anymore or more I was in love with defender and Eugene Jarvis just like my childhood hero and a couple years back I got to give him his Lifetime Achievement Award at the award ceremony as great mama thank you well Oh just as a you know playing it a lot I would always wonder to see a reverse level was something that was a happy accident or if it was something you plan from the beginning I list them as really seat-of-the-pants is we had we did level two first and then we did level 1 and

then we progress through the game when I got off time to level 5 it's like okay they've seen all that stuff how can we break the rules and I'd had a very relaxed weekend up in Mendocino or something and I thought hey yeah go uphill not a very sophisticated thought process no so a related question you know how much tuning did you have to do on the levels after you made that initial prototypes and put it in the game and did you mostly get it right from the start or did you find you have to do a lot of tuning oh yeah it was really cumbersome process to create the artwork for this so if we look at the second level the blue level you can see five rectangles at the top which are sort of your opening shot of the level is pretty much every night for a week I tried some different graphics to make try to make that pop better until Bob told me to just stop it and get on to

the next part of the game I mean very even so much easier if we'd have gone with the original plan of the eyebrow blood style hardware we just typed it in and seen it live yeah but but other things like you know the I know I could imagine like the slope on the the you know lines going down or you know like the height of some of those bumps that you would have to go over you know is that stuff that need to be tuned or was that kind of mostly got it right from the starters um in general there wasn't a whole lot of finesse to it beyond the width of the play field and if you look at it you've got one square to square or three squares and two squares seems to be about the right spot so I know it wasn't particularly tricky more the questions what was the stick going to be so why is that level going to be artistically different or what is the obstacle in your way and the actual geometry of the playfield doesn't figure into it too much all right thanks how long after the

Bennett after you finished the arcade version of marma Manas did they come out with the version for the Amiga so we finished the arcade version in December of 1984 and the Amiga version came out towards the end of 1985 I took them roughly as Long's and convert it to the Amiga as it did for us to make it in the first place because they weren't using you know the mighty power of the seven megahertz 65,000 and so on ironically a EA was heavy-duty into their electronic artist program so if you open that up the two guys who did the conversion are the featured artists and other one more of these I want to be better next to your front please come and give a post-mortem about Crystal Castles and talk about your experiences EA had all asked for permission to feature myself and Bob Flanagan and Sam Comstock in Marble

Madness for the Amiga and it sold 500,000 or a million units and tarnishment refused saying that it was a corporate product and that they did not want any of their people features in that way thank you all right well thank you very much for coming [Applause]