Is a coin flip app actually random?
You press the button four times and get yes, yes, yes, yes. It feels broken. Something in the app must be stuck, or weighted, or just showing you the same answer to save effort.
Almost certainly not. What you are looking at is what randomness genuinely looks like from close up, and the intuition that says otherwise is one of the best-documented errors in how people reason about chance.
Play Yes or No - Decision Maker
Get a straight YES or NO answer when you can't decide.
Four in a row is completely ordinary
Work it out. Each flip is independent with an even chance either way, so any specific sequence of four flips has a probability of one in sixteen. There are sixteen possible sequences of four, and exactly two of them are four-of-a-kind — all heads or all tails.
So in any four consecutive flips, you will see a run of four about one time in eight. Flip twenty times in a session and a run of four somewhere in there is more likely than not.
The reason it feels wrong is that we compare it against a mental image of randomness that is far too tidy. Asked to write down a “random” sequence of fifty coin flips, people produce something that alternates much more than a real coin does and almost never contains a run of five. Real sequences are streakier than invented ones — reliably enough that it is a standard classroom demonstration to tell the fabricated list from the genuine one at a glance.
A coin has no memory
The follow-on belief is that after several heads, tails is “due”. This is the gambler’s fallacy, and it is wrong in a specific way worth being precise about.
The coin has no record of what it did before. The next flip is still an even chance regardless of what preceded it. There is no mechanism by which previous results could influence the next one — not in a physical coin, and not in an app.
What confuses people is that long-run averages really do converge. Flip ten thousand times and you will land close to fifty percent. But that happens by dilution, not by correction: the early streak is not cancelled out, it is simply swamped by the volume of later flips. Nothing pushes back toward the average. It just stops mattering.
Why an app that fixed this would be the rigged one
Here is the part that inverts the original suspicion.
Imagine an app that noticed you had three yeses in a row and quietly nudged the fourth toward no, to make the results feel fairer. That app is no longer random. It has become predictable — knowing the last three answers would tell you something about the next one, which is precisely the property randomness is supposed to lack.
Some products do this deliberately, and it is not always dishonest. Music services have shuffled playlists that deliberately avoid playing two tracks by the same artist together, because users complained that a genuinely random shuffle “wasn’t random”. Some games apply hidden weighting so a run of bad luck does not drive players away. But in every one of those cases, the version people call fair is the one that has been interfered with.
For a decision maker, the interference would defeat the point. If the answer can be anticipated from the previous ones, it is no longer external to you, and being external to you is the entire function.
How the number is actually generated
Software does not flip anything. It uses a pseudorandom number generator: an algorithm that starts from a seed value and produces a sequence which passes statistical tests for randomness while being entirely determined by that seed.
“Pseudo” sounds like a weakness and is not, for this purpose. A good generator’s output has no detectable pattern, no bias toward either outcome, and no correlation between one value and the next. Modern devices also mix in genuinely unpredictable inputs — timing jitter, hardware noise — when seeding.
The distinction matters for cryptography, where an attacker predicting your next value is a real threat. For deciding whether to have pasta, any generator on any phone from the last twenty years is overwhelmingly more even-handed than a physical coin, which is measurably biased by which face starts upward and by how it is caught.
What the app is actually for
All of which is a long way of saying the randomness is fine. The more interesting question is why handing a decision to a random process helps at all, given that it cannot possibly know anything about your situation.
Two reasons, and neither has anything to do with the quality of the answer.
It ends the deliberation. Some decisions cost more to make than the difference between the options is worth. Twenty minutes spent choosing between two acceptable dinners is twenty minutes gone regardless of which one you eat. An arbitrary tie-break is not a worse decision — it is the same decision, made sooner.
It moves the choice outside the group. This is the real function when other people are involved. Somebody has to decide who goes first, who pays, who makes the phone call — and whoever decides carries it. Handing it to a coin, a wheel or a finger picker removes the blame entirely. Nobody chose. That is why a visibly fair method beats one person saying “I’ll pick”, even when the one person would have picked fairly.
The trick that makes it useful for real decisions
There is a well-known use for a coin flip that has nothing to do with the result.
Flip it, look at the answer, and notice how you feel. If you are relieved, that was your answer. If you are disappointed and immediately want two out of three, that was also your answer — you just had to see the alternative before you could recognise it.
That works because the flip forces a concrete outcome into view, and preferences are much easier to detect against something specific than in the abstract. It is a genuinely useful technique, and it is worth being clear that it is a way of consulting yourself, not a way of outsourcing something that matters. Anything with real consequences deserves the deliberation.
Where the app fits
Yes or No does the trivial version well: press DECIDE and get one or the other, with no question to type in and nothing tracked between presses. It also carries a coin flip, a decision wheel for more than two options, a random number generator, spin the bottle, and a finger picker where everyone holds a finger on the screen and one is chosen — the last being the fastest visibly fair way to pick a person in a room.
Nothing in it evens out the results, remembers what it said last time, or nudges you toward the answer it thinks you want. Four yeses in a row is a coin behaving normally.
Yes or No - Decision Maker
A straight yes or no, plus a coin flip, a wheel, a number generator and a finger picker. Free, offline, on iPhone and Android.
