“Atari Was Very, Very Hard” – Nolan Bushnell on Atari, 50 Years Later

I actually fixed one of those games once. It was really easy since both tanks had duplicate hardware, so signals could be compared between both tanks.

I always wondered if companies set up fake competitors, both for internal (rnd) and external (public perception) effects.

There's this remarkable Pong simulator working at a circuit level:

https://www.falstad.com/pong/clkgen.html

It describes how all the subsystem works (horizontal ball movement, drawing the score, etc.) and allows to play with them in the real time.

I dug up the citations for Kee:

> In October of 1973, Bushnell decided to grab as much market share as possible by signing exclusive contracts with distributors in each geographic area to buy only Atari games. Because most geographic areas had two distributors, Bushnell separately (and semi-secretly) created Kee Games, named after Bushnell friend Joseph Keenan who became president of the company. Kee would sign exclusive contracts with the second distributor in a geographic area. The games that Kee and Atari produced individually were eventually released by both companies with unique names and some cosmetic differences.

> Spike Was Kee’s copy of Atari’s Rebound. Kee’s products were mostly copies of Atari’s games with innovative features added to differentiate them from their Atari cousins, and create the feeling of a “rivalry” between the two companies.

> Tank was the game that saved Atari from bankruptcy in 1974. The game became so popular that the exclusivity agreements Atari demanded from distributors were thrown out the window. Everybody wanted it, and Atari made sure they got it. Bushnell’s cash flow problems at Atari were suddenly reversed. Atari and Kee merged at the end of 1974.

https://www.gamedeveloper.com/business/the-history-of-atari-...

My mom worked for a company that made chips inside the Atari cartridges and so we got a free VCS/2600 and a bunch of carts when they came out. Tank was, hands down, the best game! My brother and I spent hours really interacting with each other over the incredibly simple, blocky graphics. The "tanks" were made up of about 5 giant pixels. That's it! The planes were even worse and didn't look like planes at all. We had more fun competing and really playing with those terrible graphics.

I wonder if all the high-end game qualities today take away some of that genuine player interaction? I do love modern games and I doubt I'd enjoy Tank today, but it was special in its time.

How is that different than Nintendo?

Edit: Other than obviously coming before Nintendo

w.r.t the "good part", there is a difference between games as toys and games as a story.

a lot of games today are "story games", even if there isn't a fixed story, there is some progression and different parts of the game open up at different times and result in different emergent gameplay. even something sandbox, toy-like like cities skyline and minecraft has a progression.

i am reminded of gta V, which i thought had a really good story, where the game jumps you into a bank heist in the first scene of the game, it was a huge scenario with lots of set pieces, a lot of mechanics are introduced right away, it's pretty intense, and thwarts the RPG baby-crab-killing trope, thank goodness.

games as toys, like pong i guess, where you begin the game immediately having access to every single feature of the game, still exist - apex legends, cs go, rocket league. but you still won't be able to jump into it at a decent competitive level right away. in most popular competitive FPS games, you will be rocked for a good first 20-50 hours of gameplay until you build understanding.

w.r.t to what nolan means when he has to go through an hour of things before he can start playing the real game, sadly that's a historical era. games today have so much more features, so much more complications, a mix of meta thinking, hand-eye coordination, and mechanical practice that it's hard to simply jump into a game and be comfortable right away. a lot of games today need some amount of practice, whether 10 minutes, 10 hours or 10 months.

I'm the same. May be biased, as I sell retro games for a living, but I really miss the era where you could just chuck in the disc/cart and play.

PlayStation and Xbox are even worse in this regard. Not only do you have the long intro, but a solid half-hour of installation and patching before you can even launch the game!

On the flipside, my formative video game experiences involved reading the instruction booklet to a new NES game, cover to cover, while being driven home from Toys R Us. They were essential to knowing how to play and appreciating the storyline.

The Atari 2600 was no different: You needed the manual to know what the weird symbols onscreen meant, or what the switch settings would do for that particular game.

Of course, things were different in arcades.

Agreed. Contra is my go to when I think of this experience. Just dropped into a war zone and you start shooting. No explanation necessary.

Depends on the game. If you pick up a strategy game you can just jump in just fine. If you want to play Counter-Strike, just start up, join server and start shooting.

same. as I get older....I want story...but not at the beginning...and I don't want a tutorial. I want gameplay...with cutscenes like breaks...the cutscenes should be skippable and pausable.

as for Multiplayer....Less loading screens, lobbies and loot crates.....and maybe just maybe there is hope.

For what it’s worth I also remember hearing that the Jaguar was incredibly buggy. If they had more time they probably could have corrected that so developers didn’t have to worry about it.

The Saturn was the one that had SH2 for a main CPU and a 68000 for a sound chip (the predecessor Genesis had a 68000 as main CPU and Z80 as a sound driver)

Jaguar had two weird RISC chips (Tom and Jerry) meant to be used as DSPs, and one had a few more instructions for sound, and the other had something a teensy special for graphics. It had a 68000 that was meant to be used as a "manager" and "read the joysticks". To top it off you had the actual display chip (the "Object Processor") which was weird? flexible and programmable in some ways that almost seemed to me to hearken back to Atari's ANTIC chip in its 8-bit line.

The 68000 was the only "normal" thing about that whole system.

"This is the heart of the game. I wanted each ghostly enemy to have a specific character and its own particular movements, so they weren't all just chasing after Pac Man in single file, which would have been tiresome and flat." — Toru Iwatani, Pac-Man creator

https://gameinternals.com/understanding-pac-man-ghost-behavi...

I can't answer for Atari but I know that games from that time would usually derive randomness from user input timings. There simply wouldn't be a readily available source of (good) entropy available otherwise and using a complicated algorithm to harvest it and "mix" it would be prohibitively expensive. And who cares for crypto-grade entropy in a game anyway?

It's actually still common with much more recent games where there are viable strategies that involve putting the RNG in a known-state (for instance at init) and then performing very precise actions to trigger a normally random event with 100% certainty. Here's an example for Final Fantasy XII: http://www.fftogether.com/forum/index.php?topic=2778

Basically the strategy is that by using a "cure" spell several times in a row and looking at how much health it regenerates in the game you can guess the position within the RNG output sequence (that is fixed and resets when you power down the console). Then when you know where you are you can plan your actions knowing ahead of time what the RNG will output and whether it'll be favourable or not.

I believe that what you observed in Pacman was however due to a primitive AI: each ghost would have a simple strategy and stick to it, making them predictable and non-random by design.

The ghost patterns in most Pac-Man gameplay are designed on purpose, but the directions they turn when running from Pac-Man once he’s eaten a pellet are intended to be random. From a broadly interesting look at the game’s internals (https://www.gamedeveloper.com/design/the-pac-man-dossier):

“The PRNG generates an pseudo-random memory address to read the last few bits from. These bits are translated into the direction a frightened ghost must first try. If a wall blocks the chosen direction, the ghost then attempts the remaining directions in this order: up, left, down, and right, until a passable direction is found. The PRNG gets reset with an identical seed value every new level and every new life, causing predictable results.”

I don’t know how the initial RNG works (the source is available online, but my Atari ASM skills seem to be lacking), but more modern takes sometimes use a linear-feedback shift register, as seen here: https://www.randomterrain.com/atari-2600-lets-make-a-game-sp...

Each ghost was designed to be more or less aggressive based on their colour.

Red was always coming straight for you and you better be watching for that guy!

Pacman uses a pseudo-random-generator in the frightened-ghost movement (to pick a random direction at intersections - but only for ghosts in frightened mode).

In general "randomness" and "balancing" don't work too well together.

PRNGs in arcade machines and home computers often used a variant of xorshift since that's easy to implement in hardware, but most often this was just for noise generation in the audio chip.

Tempest is a very difficult game after the first 8 levels or so. In a good way.

My father and I in the early 90s were lucky enough to come across a Tempest arcade machine that "didn't work" in a yard sale and bought it for 50$. I was going to make something out of the cabinet. I opened it up, and noticed the CRT was disconnected from the big white input on the center. Connected it, plugged it into a surge protector waiting for smoke... and had a working Tempest game in my room for my teenage years.

Always a fun game, and when I learned you could change game settings with DIP switches, that curiosity never really left and led me to coding. 30 years later, I'm still coding.

Whenever people complain about the difficulty of Fromsoft games, I just think back to the early days of Atari and the pre-battery-save NES. No continues, you just go as far as you can and if that's not good enough, you get to start from the beginning until you get past it.

I remember having pole position as a cartridge for my Atari 800XL, plug it in, turn it on, and instantly into the game! I played that game a LOT.

Mobile games today seem to have Atari’s “easy to learn, difficult to master” design philosophy.

Bushnell suggests the opposite near the top of this interview, that part of a coin-op game was exactly that you did need to let people plop a quarter in and just start playing!

> Nolan Bushnell: A little bit. Remember that Atari was founded as a coin-op company. And coin-op has this requirement that a newbie has to get into the game almost instantly without reading instructions. So the simplicity of onboarding is lost by a lot of people right now.

Which makes sense to me, of COURSE you have to be able to "just start playing" if you wanted to get someone to put a quarter into a machine they hadn't played before!

(nitpick: Pole Position was created by Namco. Atari was the distributor in the USA)

It is just the opposite. Haha their when they are easy and anyone can play them. It is why there is such a large mobile game market today. Pole position wasn't hard and didn't take to long to learn. You can't compare it to a game with superior graphics and controls. For the hardware pole position is pretty good.

Having a fire fast culture creates a stressful atmosphere.

It's better to have high standards to begin with to avoid needing to fire people. If you're firing a double digit percent of your hires, there's something wrong with your hiring process.

Most companies use metrics that are weakly correlated with success on the job. When you've hired dozens of people, who will be successful or not becomes quite obvious.

> I feel two ways about this. You need to give people time to grow into a role and time to get really good at it. But you also need to watch closely and make sure that growth and expertise are occurring, and if not, get better people in.

I found the way you chose to frame this to be a bit disturbing. You framed the employer as a passive watcher of a process where they invest zero and provides zero input, except for the part where they feel entitled to terminate staff for the sole sin of not blossoming greatness in the vast desert of growth that was your creation.

How about this: if what you seek is growth and expertise, why not explicitly nurture that in the environment you create?

Atari treated their employees poorly, which is why Activision was formed. I give Bushnell zero credibility on this point.

In fact, it could be argued that Bushnell’s poor treatment of employees was the harbinger of the first video game crash.

I mostly like shooters (like Far Cry) and skip every cut scene I can. I wish games had an auto-skip all cutscenes option. It drives me nuts to have to skip more than one cutscene in a row. Do they really think I’m going to want to skip three cutscenes and then watch the fourth?

I wish game intellect was far more advanced than graphics, by this point. Without good in-game AI and unless there's really good writing (which is rare on itself these days!), it all gets boring really fast. Talking about single-player here, of course.

That's an entry level FPGA challenge:

https://www.fpga4fun.com/PongGame.html