Pac-Man & Scoring

Some thoughts on Pac-Man and high scores.

There are a few omissions and simplifications for the sake of clarity.

The optimal path through Championship Edition stages are not always as obvious as they appear to be for a number of reasons, although I touch on this when I say sometimes not scoring short term can lead to a better score long term.

Also Pac-Man can pre-turn in both games which makes him round the corner faster so his movement is a little more complicated than it appears to be, the speed of championship edition makes it much harder to pre-turn for every corner but it's far from impossible.

I'm sure not every version of Breakout is completely deterministic (often the ball drop position is random) but the game can certainly be built that way and has been on many occasions.

[PhluppeHimself]

This guy could review paint drying for 20 minutes and I would still watch it.

[AndyNPC]

Here's an interesting example of a game achieving skill based scores: Spelunky's Daily Challenge Mode. Normally Spelunky is a game where the levels are generated procedurally based on a random seed. There is a limited amount of treasure to collect each game, and that is further limited by what resources the game generates, so players already are on uneven ground just by having a different seed.
The daily challenge however gives all the players the same seed and allows them only one attempt at it. This means that everyone has the same potential maximum score, but since nobody has the opportunity to practice that particular seed (outside of hacking) nobody can create a perfect strategy that achieves the top score every time.

[ashthedood]

Would love more random observations like this. Reviews are great but this really brought up a great point. 

I like pac-man, it's fun but if you can just re-peat the best pattern with no change to always get the top score then what's the point in competing.

[FSSZilla]

When the rubric of scoring elements was shown, I immediately thought of Super Hexagon as one that meets the dynamic-endless criteria.

[coreypnorris]

Two videos in two days? You're really spoiling us Matthew.

[sethmoyes]

Without a doubt this is the best content I've viewed on YouTube. The amount of time and effort poured into each video is self evident.
Subscribed.

[Trinexx]

8:05 a good example for the upper right is Black ops zombies. Once you hit about round 20 its just running in circles. And by about round 35 only the Thundergun/Wunderwaffe/other weapon with infinite damage is effective. So its just running in circles over and over until your game crashes at about round 170-200. Still find it fun though.

[TheFantasticJoe]

This was a very insightful video that got me to think of scoring systems in games in a way I hadn't quite considered before. Nicely done.

One minor note - the game you primarily show is actually Pac-Man Championship Edition DX, which is different from the original Championship Edition as it includes multiple chains of ghosts, bombs, and slowdown when approaching ghosts.

[MelftheAdventurer]

This was a fascinating video! I've never really considered the way scoring systems would influence competitive play for arcade games, despite being deeply ingrained in competitive gaming in general. Without facing another player directly, the scoring system has to be much more carefully reasoned than I would have imagined.

[michielkroder]

Just wanted to add that in Championship Edition DX (I believe that's the game you were actually referring to), there is an additional dimension to the scoring besides following the optimal path; it's figuring out when to eat the power pellets + when to hold off (as to make the huge clusters of sleeping green ghosts appear) and when to eat additional power pellets to maintain your powered up state in order to keep your combo going. Granted, it can be seen as an extension of following the optimal path, but it's a far less obvious side to scoring optimally. Fantastic video!

[adeltgm]

this was a great analysis. i've always thought a lot about scoring systems and how to improve them and you seem to share a lot of my own sentiments.

while i understand the flaw in limited/timing based scoring systems, though, i still find them preferable a lot of the time. with an endless game, as players continue to get better and better, you'll eventually reach a point where going for a high score is more of a battle of free time and patience than of skill, as the time required to compete becomes more and more inconvenient. even with the issue of limited games creating a maximum score, i think if your game is designed well enough, and the skill ceiling and potential score high enough, you'll never have to worry about that happening within the lifetime of any player.

some good examples i think are the Tetris the Grandmaster games, especially the more difficult modes in TGM2 and 3. a fully-optimized run of Death, Shirase, or Ti Master would require the absolute perfect combination of finger precision, forward thinking, and a bit of luck. it's hard to imagine it ever happening; i couldn't give you an exact chance but i'd imagine there are a lot of zeros after the decimal point.

[RupeeClock]

When considering deterministic game design and human limitations, it's important to factor in possibilities illuminated by tool assisted play, where human reaction times and errors are removed leaving only perfectly executed strategies.

I am sure that the hypothetical scenario of a perfect score in a perfect time has been achieved in tool assisted Pac-man, and if not, can be done.

[GurdevSeepersaud]

It's like I've had this thought in my head for years and years but only now I'm hearing someone who can eloquently put it into words lol

[MrTwrule]

Your discussion of ideal future inputs got me thinking about what the ideal input scheme would really be. I assume when you mentioned 'neural' inputs in contradistinction to tactile controllers, the idea was that it would remove all obstacles between that player's desire to act in a particular way in the virtual environment and that action being carried through. I think there's something right about that, in that the ideal is to create such 'transparency' of control - in other words, the ideal control scheme is to have that scheme create a perfect conduit for the player's will to act on the game world.

I would add that if when we thought of this ideal scenario, we pictured a player sitting in front of a monitor with his hands empty but somehow controlling the game purely through thought, this actually would *not* create that ideal transparency. This is because human beings are constructed in such a way that we are accustomed to navigating our own world in a rather tactile way, due to how we are embodied - so much so that it would be jarring and awkward for us to try to navigate a virtual environment without making inputs in any analogous tactile form (for example, pressing buttons on a traditional controller).

Even motion controls, if that technology were perfected, wouldn't be as effective as a traditional controller in this regard, since we are also used to getting feedback from our environment directly as we interact with it in tactile ways (not to flailing through empty air that provides no resistance, or walking in place, etc.); in-game feedback to button presses plus a rumble feature on a traditional controller can better meet this requirement. This latter point holds particularly if the game-world itself is meant to be immersive (unlike something like Pac-Man), but even playing something like Pac-man without any kind of tactile contol would be awkward in instances where the game did provide feedback meant to be satisfying in a vaguely tactile sense (like when the game pauses for a few frames and plays a distinct sound effect whenever Pac-man eats a ghost). Of course, new control schemes would encourage new ways of designing games around it, but its difficult to see how those new games would fill in that void of tactile responsiveness going both ways (to and from the player).

There may be better ways than we've thought of so far to perfect traditional tactile controls when used in conjunction with a monitor, but probably the true ideal control scheme would be full immersion VR environments (ala Star Trek Holodecks), which actually allowed us to navigate virtual environments using our own bodies and thus with the same embodied coping skills we use to navigate the real world (assuming this included some way to integrate realistic tactile feedback from the virtual objects/environment we encountered). But maybe this is what you had in mind.

[TeRHyP]

oh my fucking god 2 new matthewmatosis videos in 24 hours

i'm shitting myself in pure happiness

[TotallyNotEwan]

You can hear him struggling at 1:49, desperately trying not to say "tree" instead of "three"

[carljonsson]

Man you make me so happy with these new videos! I keep rewatching your old ones all the time. 
Keep it up! we luv u.

[mayordump]

As somebody who makes arcade-style games in his free time, this break down of dynamic challenge & endless approach is super helpful. thanks

[Parivertis]

It's funny, I spent a good long time thinking about exactly this during the initial design phases of our currently in development game 'VIVID' (mobile arcade game, with Rez & Super Hexagon inspired aesthetic and sound).

That balance of dynamism or randomness vs live player skill within a simple set of systems is probably the hardest thing to get right, especially with regards to spawn locations. After all, you don't want the player to feel like they were cheated out of a highscore due to a totally unforeseeable and unavoidable series of circumstances.

[WaddleDee]

Really intelligent observation you made. Your videos are often eye-openers for either whole games or certain mechanics.