Classics and chance is the corner of gaming where old games meet randomness: the 1958 lab demo that started the medium, the speedrun categories that turn a fixed game into a gamble, and the esports scenes that grew out of both. The practical question underneath all of it is whether the randomness is the point or the obstacle. A slot machine is chance with no skill; a chess clock is skill with no chance; most of what people actually love sits in the messy middle, and knowing which side of that line a given game falls on tells you how to practice it, how to watch it, and how to bet your evening on it.
Tennis for Two is the origin story, and it is stranger than the legend
Tennis for Two was built in 1958 by William Higinbotham at Brookhaven National Laboratory, and academic and historical sources cite it as one of the first video games designed purely for public entertainment. It ran on an oscilloscope, not a television, and the display measured 5 inches across. That is the number worth holding onto: the first widely credited entertainment video game was played on a screen smaller than a modern phone, in a government lab, by people who were not trying to invent an industry.
The design details matter more than the trivia. Tennis for Two was a side-on tennis match with a bouncing ball and a net, controlled by boxes with a button and a dial, and it drew a crowd at the lab's open house. It was not a product. It was not patented as a game. It existed to make a tour of a nuclear research facility less boring, and it worked. When you play a modern tennis or pong-like game, you are using an interface that was solved in principle before most of the people reading this were born.
The reason this belongs in a piece about chance is that Tennis for Two had almost none. The ball trajectory was deterministic given the same input and timing; the skill was in the dial. That is the ancestral model for competitive gaming, and it is the opposite of what a lot of modern "classics" collections sell. If you are buying a retro compilation expecting the purity of the 1958 model, check what you are actually getting, because most of what gets labelled classic in 2026 is arcade-era design with heavy randomness baked in.
Chance in old games is a design decision, not a flaw
Randomness in classic games usually does one of three jobs: it keeps a fixed board from being solvable, it compresses skill gaps so weaker players stay in the match, or it manufactures replay value by making each run different. Pac-Man's ghost behaviour is deterministic but opaque; a deck of cards in a digital solitaire is genuinely random; a roguelike's item drops are random with a seeded structure underneath. Treat these as different problems. The first rewards study, the second rewards nothing but patience, and the third rewards understanding the seed system.
The cost of randomness is that it makes practice harder to measure. In a fully deterministic game, a failed run is your fault and you can review the tape. In a high-variance game, a failed run might be a bad draw, and the tape tells you less. Players who get good at high-variance classics learn to separate decision quality from outcome, which is uncomfortable and slow. Players who do not learn this tend to blame the game, and sometimes they are right.
Where randomness goes wrong is when it is hidden. A game that shows you the dice lets you plan around them; a game that rolls behind the screen and only shows the result is asking you to trust it, and old games were often not built to be trusted. If a classic title feels unfair in 2026, the question to ask is whether the randomness is visible. Visible variance is a strategy layer. Invisible variance is a slot machine wearing a costume.
Speedrunning turns a fixed game into a chance problem, and the categories are the tell
Speedrun categories are the clearest modern expression of classics and chance, because the category rules decide how much randomness you are allowed to remove. Any% lets you skip content and often lets you manipulate random number generation; glitchless categories ban the manipulation and leave you at the mercy of drops and enemy patterns; 100% categories force you to engage with the parts of the game the designers randomised on purpose. The category name is not housekeeping. It is the contract that says what kind of gamble you are taking.
What goes wrong for newcomers is entering a category without reading the ruleset, then wondering why their route does not match the world record. Route documentation for a heavily randomised category will include backup lines, because the runner expects the good outcome not to happen. If you watch a record run and it looks effortless, you are watching the one attempt in many where the dice cooperated, and the runner's skill is partly in recognising that moment early and not wasting it.
The trade-off is time. Low-variance categories are easier to grind and easier to compare; high-variance categories are more exciting to watch and much slower to improve at, because your practice sessions do not produce clean feedback. If you have limited hours, pick a category where the randomness lives in places you can route around. If you have unlimited hours and a tolerance for frustration, the chaotic categories are where the interesting routing work happens.
Playing old games today means choosing an emulator, a display, and a definition of authentic
There is no single right way to play a classic in 2026, but there is a right way for each goal, and the goals conflict. Accuracy-focused emulation tries to reproduce the original hardware timing, which matters for input lag and for games where frame-level behaviour is part of the design. Convenience-focused emulation adds save states, rewind, and fast-forward, which change the game into something easier and arguably better for a first playthrough. Original hardware is the reference point and the most expensive option, and it comes with the display problem: a modern flat panel does not handle the signal the way a CRT did.
The display question is where most people get surprised. Old games were designed around the latency and blur of the screens they shipped with, and moving them to a flat panel changes how they feel before you touch a setting. If a classic feels wrong and you cannot say why, the screen is the first suspect, not your skill. The second suspect is the controller, because a modern analogue stick is a poor substitute for a digital pad in games built around eight directions.
The honest trade-off is that chasing authenticity costs money and space, and most players do not need it. Save states are a legitimate way to finish a game you would otherwise abandon. The one place accuracy genuinely matters is competitive or comparative play, where a timing difference between setups makes results meaningless. If you are chasing records, match the setup to the category rules. If you are chasing the experience, use the conveniences and do not apologise for them.
Esports grew out of this material, and the modern scene has rules the classics did not
Esports is organised competitive video gaming, typically played in leagues or tournaments with structured rules, teams or individual players, and often professional-level prizes and sponsorships. That definition is worth quoting because it explains what separates a competitive scene from a leaderboard: structure. A leaderboard is a list. A league has eligibility, scheduling, officiating, and a ruling body that can disqualify you. The moment a classic game gets a competitive circuit, the randomness question stops being philosophical and becomes a rules committee problem.
That is why competitive rulesets for old games tend to ban or constrain the random elements. If a tournament result can be decided by a drop table, the tournament is not measuring skill in a way sponsors can sell, and sponsors are the reason the prize pool exists. The cost of this is that competitive play drifts away from what made the game fun casually. A game that is delightful because of chaos becomes, at the top level, a game about controlling chaos, and the two experiences are different enough that fans of one often dislike the other.
Broadcasting changed the shape of this. Twitch is a live-streaming platform focused primarily on video game and esports content, where users can broadcast gameplay, watch other players' streams, and participate in real-time chat and community features. The minimum age to have a Twitch account and stream is 13 years, which sounds like a technicality until you notice that it sets the floor for who can build a competitive reputation in public. A scene that runs on broadcast exposure is a scene where the youngest competitors can be visible before they can be paid directly, and that gap is a persistent structural issue in competitive gaming.
The other cost is that broadcast-friendly games get supported and broadcast-hostile games do not. A slow, thoughtful classic with long decision times and no obvious spectacle will not attract a streaming audience, which means it will not attract a circuit, which means its competitive scene stays small and informal. That is not a failure of the game. It is a mismatch between the game's pace and the format that pays for competition.
How to decide what to spend your time on, in order
Start by deciding whether you want to play, watch, or compete, because the three have different randomness tolerances. Playing rewards a game whose variance you find interesting; watching rewards a game whose variance produces drama quickly; competing rewards a game whose variance is small enough that the better player usually wins. Most people want two of the three and end up disappointed because they picked a game that only delivers one.
Then check the ruleset before the game. For a classic, that means knowing whether your version is the original release, a port, or a remaster, because ports change timing and sometimes change the random number generation. For a competitive category, it means reading the category rules and the tournament rules separately, since they often disagree about what counts as a legitimate technique. For a broadcast, it means knowing whether the stream is live or a highlight, because a highlight removes exactly the variance that made the match interesting.
Then set a budget in hours, not money, because the expensive failure mode in classics is buying hardware to solve a problem that is actually about practice. A better display will not fix a route you have not learned. A more accurate emulator will not fix a category you have not read. The cheap interventions, reading the rules, watching a full run rather than a highlight, and playing the version the community actually uses, do most of the work.
The limit of all this advice is that classics and chance is not one hobby. It is a shelf of overlapping hobbies that share a history and disagree about what matters. The person restoring original hardware and the person grinding a random-heavy speedrun category are both working on the legacy of a 5-inch oscilloscope demo, and they would probably not enjoy each other's weekends. That is fine. The useful thing is to know which one you are, because the decisions that follow, about setup, category, and competition, are only coherent once you have picked a side of the randomness line.
Deterministic classics punish practice differently than random ones
A fixed game is harder to blame and easier to coach. When the same input produces the same outcome every run, a missed jump is a timing error you can isolate, timestamp, and drill; the feedback loop is short and honest. That is the Tennis for Two inheritance, and it is why so many old arcade games are still studied frame by frame. The cost is boredom: a game with no variance eventually stops surprising you, and the ceiling arrives before the interest does.
Randomised classics invert the problem. You improve in bursts, and some sessions give you nothing to learn from because the run never reached the part you were practising. Route around this by setting up controlled starts, save states that drop you at the hard section repeatedly, rather than replaying the opening each time. The trade-off is that states trained practice can mask the fatigue and resource management of a real run.
Version matters more than genre when you go looking for a classic
Before you commit hours to an old game, confirm which release you are playing, because the original, the port, and the remaster are three different games with three different timings. Ports built for different hardware often shift input response and, in some cases, the random number generation, which quietly invalidates any route or strategy written for the original. Community wikis usually note which version a technique was verified on.
If you intend to compare your results to anyone else's, the version is not a preference, it is part of the ruleset. Match it, or accept that your numbers are yours alone.



