How come I don't see much data-mining for pinball?


To me “data-mining” is extracting statistical metrics off something. And it was a joked comment on my behalf. What we have here is “code analysis” or “reverse engineering”.

You know. I have thought a lot about this. And I am reluctant to do it. And that honestly includes finding out stuff and keeping it to myself. Some pinball myths are best left to live.

Unless it is bugs or semi-bugs* (obscene random awards etc.) or something that I am super curious about myself, I am not that interested.

(*) Or we can name them “competition-bugs”.


I think to answer the big-picture question in this thread, it comes down to this:
Pokémon (and other games listed) are games of tactics and strategy. Pinball is a game of skill.
Sure, pinball requires strategy too, but if you don’t have the skill to hit the shots you want, you could have the best strategy in the world and still suck. Pokémon doesnt have that element. Once that element of skill and reflexes is gone, your strategy becomes way more important, and an encyclopedic knowledge of rules becomes much more important, because what else do you have?

What I’m suddenly curious to see is if you could build a machine to hit every shot you wanted guaranteed, what would happen if you gave everyone three minutes to a game and had that kind of competition. I suspect suddenly obsessing over every little detail would become the norm in that format.


Glass-off two-minute drill? That would be amusing.

You might have to impose some limits, though, like you have to throw the ball where you want it to go, not just touch switches by hand.


Tool-assisted speed runs in pinball could be cool to try and find the theoretical maximums


computer vision controlling flippers would be super cool.


This could also be done with the IR sensors on the multimorphic system. It would be awesome to train a computer to play and have a “deep blue” style showdown between the computer and top players.



I have some pretty mixed feelings on the patent system as a whole, but it’s kinda cool to see the system “working as intended” with all these 90s patents expiring, opening the doors for fresh implementations.

Less than a year until the final Williams patents expire, according to Justia.


The Phantom Flip system, for all its faults, is actually very very smart, and you could do a lot worse than to start from there and work towards something better.


Is that what Thing Flips on TAF uses? Or isn’t there something called Phantom Flip on Monster Bash? I’m usually too busy bricking the Frank shot to pay much attention.


Baywatch also has a game controlled flipper.


Thing flips uses the targets on either side of the swap to calibrate its (single) shot. (Above the swamp, too early, below the swamp too late). It gets better all the time.

Monster Bash takes this idea a bit farther. There’s 2 magnetic eddy sensors in each inlane, the software measures the time it takes for the ball to go from one to the next deriving the speed from it, which then times the shot (if you don’t press the flipper button cancelling it). It is a very cool feature that again becomes more accurate over time (based on the misses).

Unfortunately, I don’t trust it especially on location machines if the sensors aren’t working properly you can just get a ball draining off of it. If I owned a monster bash I’d probably disable it just for consistency’s sake - in competition I’d want to be shooting the shot I want instead of letting the machine decide.


All right then. Is this the majority opinion among pinball fans? (I can say that it is not the majority opinion in any circle of video game fans I’ve been in–in general, they want the answers, and they want it as soon as possible.)

On the other hand, data-mining is also quite frequent in games with a strong component of non-strategic skill as well. I used Pokémon as an example because it’s a franchise I know very well, and its nature as a numbers game makes it easy to explain. But I’ll use another example, in a genre that gets picked apart and disassembled just as much as Pokémon. This is a game that just came out this week, BlazBlue: Cross Tag Battle. It is a 2-D fighting game. Think Street Fighter or Mortal Kombat if you’re not familiar with more recent franchises in this genre. Perfect timing and twitch reflexes are essential to being decent at games like these.

(Admittedly, I chose this screenshot from my recordings because Weiss Schnee is in it.)

Well, the Steam release has already been data-mined, and some of the characters meant to be released later have been made playabe by these data-miners who figured out how to unlock them without paying for them and who explained how to do it.

That is, a general rule among video games with a competitive scene, or any sort of popular multiplayer, is that someone with skill will, more often than not, be defeated by someone with equal skill AND knowledge of the game’s inner workings. Obviously, you don’t need to know much about the inner workings of BlazBlue: Cross Tag Battle to unlock the DLC characters ahead of time, but someone who mains Ragna, for instance, and knows the exact size of Carnage Scissors’s hitboxes and when the super armor frames will kick in will have an advantage against another Ragna main who doesn’t, assuming equal skill otherwise. Hitboxes (areas when a character attacks that deal damage) and hurtboxes (areas where a character is vulnerable to attacks) are some of the first things 2-D fighting game data-miners look for, because this is information normally hidden to the player but is valuable information for anyone attempting to play at the highest levels, and indeed, all of the top players for a fighting game will know them all for their own favorite characters at minimum, but preferably also the ones for the most popular characters too.

That’s why this has bugged me so much: It is no different from finding out the rules to a pinball machine, and it’s a more efficient way to do it than playing the game itself, perhaps short of removing the glass and just poking at the switches yourself.

Very few of the Addams I’ve played on location have anything close to an accurate Thing Flips. Most of the time, they undershoot the ball. Sometimes, they drop the ball into an outlane. I suspect that some of these machines don’t have the flipper strength to shoot that high up anymore, so there’s nothing the programming for Thing Flips could do.

Thing Flips would be something great to data-mine though, for curiosity’s sake. Unless that’s already been done. I know it’s patented, but consumers don’t have to worry about patents.


I think what you’re describing rarely exists in pinball. Pinball machines do not generally have DLC or time-delayed features, nor do they have major details obfuscated from the player. They may have Easter Eggs, but generally those do not represent significant scoring advantages. (e.g. I don’t think you’ll find anyone rushing to play Breakout on STTNG in a competitive setting, even if they know the magic incantation to make it happen, which was a multi-decade mystery)

The trick in pinball is that even if you know exactly what to do strategically, you have to execute that sequence in the correct order with the correct timing… i.e. be perfect with your shots. That’s much easier said than done. In fact, if a player was literally perfect with their shots, it’s kind of irrelevant if they know the rules, because they would never drain, and could just execute a low-value shot infinitely (*). So in practice, pinball becomes a matter of an X% probability of making shot Y correctly, with the assumption that shot Y is actually appropriate at the current moment given the current game state.

(*) Kinda sorta. If the shot(s) in question definitively fed to a flipper, this should be true. Shots that bounce around more chaotically are harder to assess with 100% certainty.


I guess not.

Take Casino Run, for instance. Even being random on multiple levels, the only thing I would like to know for sure, is, if the bomb is indeed never seen on the first roll. I play by that belief.

Otherwise you sort of know the risk/luck stakes from playing it. The expected payout and how to play it stratigically safe.

Casino Run is per definition always worth playing. And whether it is worth going for is more a matter of how the saucer is to shoot, the eject return and ease of relight.


Indeed–a lot of data-mining is to look for trivia or upcoming stuff, and I definitely get you there, as I know there usually isn’t a lot of dummied out data for pinball machines. (Which reminds me, as Family Guy has been brought up earlier: Does anyone know if Not Special was ever meant to do anything? It just screams out “Dummied Out Content” to me, and I’d like to know if there was any leftover data anyone found in regards to that.)

I think another big thing is that pinball programmers are much fewer in number and tend to be a lot more upfront about these things when asked…though for some reason, I see a lot of evasive answers as to the rules outright. I understand the mindset of “discover for yourself,” but I came from an environment in which rules for video games are easily and readily available, and you are expected to have memorized them for a competitive setting because you know your opponents will have done so.

But I think you’re right, overall, because even among the very best pinball players, they’re not going to play perfectly or even close to perfectly, whereas if you see the top players for any competitive video game, they hardly ever deviate from optimum play, and said deviation is usually what creates the loss (though it can just as often be an incorrect guess as to what the opponent will do, and the two parties are otherwise playing near perfectly, at least with the available metagame knowledge at the time).


To further this point, games you mentioned previously are PvP, as compared to pinball, which for lack of a better term, is PvE.
If you are against another player, knowing everything about their character/abilities/moveset etc tells you their weakness or vulnerabilities to be exploited, and you need a detailed knowedge of the game mechanics to do so to pick a counter that you can play well and not try to adapt to a new moveset on the fly (relying heavily on my limited SSB64 competition knowledge here, I’m sure my lingo is off). If I know my move has priority over theirs, that changes my strategy. And if I flub an attack, I have someone to immediately punish me for it with a counter combo that drains a health bar.
If I’m playing pinball, I don’t need to know a pop bumper is worth 10k points if I know the pops are a dangerous shot a la Aerosmith. All I need to know is what shots I can make reliably and what modes will make it worth points, and what shots to avoid because a miss is deadly. I don’t need details, because I’m not on a time limit or have someone exploiting my mistakes. Even if I brick a shot, there is no consequence right up until I drain. If I only shoot safe shots (and this varies by tournament and skill level), I can brick shots all day and not be penalized. Watch Bowen in the NYC Pinball Championship finals brick shot after shot, but recover, and still crush it (I forget which game. It was glorious though).
None of that ‘big picture’ strategy in pinball is really in the code to begin with. The code doesn’t dictate how hard a shot is, or how punishing it is to miss, so what is left to dig through? You are better served just playing the game regularly.


I find this statement difficult to understand. How are pinball developers both upfront and evasive?

Iron Maiden had the whole backstage pass with all the rules of the game on it. Is this what you’re talking about?


A lot of the responses are - there’s no point, it’s pinball. But I don’t think this addresses the question asked. Sure, we aren’t going to play perfectly, but I think with 2 people of similar skill level, a person with complete rules knowledge will win more often than a person playing with little or none. And generally speaking, rules knowledge is sought and shared on forums like this for just that reason. So why hasn’t anyone hacked the code or data mined (whichever term you’d rather use)? All it would do is inform your shot selection accurately. And maybe it would confirm what we already do, but maybe not.

Maybe it is that we’re so used to getting our knowledge by trial and error? I don’t know. Maybe we don’t like the idea of one optimal strategy? There are plenty of games where everyone already only does one or two things, because they are the only ‘legit’ tourney strategies. And maybe a data mine would eventually do this to every game? And ruin pinball.

I think what a data mine could help with is maximizing points at different stages in the game, and we could combine that maximizing with our own assessment of risk. For instance: Iron Man. Once you raise Monger, spinner rips can are worth 7.5k per spin and increase the value of hits to the monger by 7.5k per spin. You need to hit monger 6 times to start the multiball. I never see people shooting spinners prior to hitting monger. Are these points too inconsequential? The spinner shots are probably the least risk shots in the entire game, assuming there is nothing about a particular orbit shot that makes it dangerous. Each spin prior to the collection of a monger letter will be worth a bit over 50k points when all is said and done (7.5k immediately and 7.5k for each of the 6 hits to monger). In a competition where finding strategies that minimize risk and maximize returns, couldn’t this be something worth looking into?

Just knowing the option that awards the most points won’t fully inform our decisions, but it could be a jumping off point. So I also wonder why nobody has data mined pinball code.

Edit: I found the Iron Man info here
Don’t want to make it seem like I learned this or even tested it. I went looking for it when I ripped a spinner during monger mode and it seemed decent. I thought - what if I take bogey into monger and just rip the spinner all day instead of attacking the monger - and looked for info, this is the best description of what would happen.


That’s definitely true in the current production environment, and probably is true going back a few years. Discounting outliers like TNA, games being produced today are so complex with such intricate stacking strategies that you are at a massive competitive disadvantage if you don’t have the money to own one or don’t have access to one in your area. Stream archives can help, but there really isn’t any substitute for table time.