About Good Brain: what we do, where the data comes from, and what we don’t promise

What this site is

Good Brain is a free brain-training site you can open straight in a browser. It holds 11 original mini games right now: nine are live and two are still in development. Between them they cover four cognitive areas — memory, attention, reaction and logic. The playable ones are Shape Memory, Schulte Grid, Reaction Time, 24 Game, N-Back Working Memory, Bean Pick, Flash Click, Sudoku and Colour Imprint. Each is a self-contained mini game with rules simple enough to grasp at a glance.

Every level has eight difficulty tiers that unlock in order: reach the pass line on the current tier and the next one opens. Scores feed a site-wide leaderboard so you can compare your best with other players, and any tier you have unlocked can be replayed as often as you like. Scores are recorded from the server-side recomputation, so the same history follows you across devices — nothing is lost when you switch screens.

Nothing to install: it runs on phones, tablets and desktop browsers alike. You can play without signing up, too — in guest mode your scores are simply not saved and never reach the leaderboard. Sign up for an account and your scores and unlock progress are recorded. Guest mode is a good way to try the feel of it; an account is worth it if you plan to keep playing and want to watch your own progress curve over time.

How difficulty and scoring are set

There are no secret multipliers and no hidden difficulty that quietly adapts to your skill. Each tier changes one thing that can be explained in a single sentence: how wide the grid is, how long the sequence runs, how many candidate icons appear, how many categories and pieces are in play, how large the board gets. Every level page lists the exact parameters for each tier, so nobody has to guess what "harder" actually means.

Scoring does not happen in the browser. The front end reports raw measurements only — elapsed time, number of correct answers, where a piece was dropped — and the server recomputes the score with the same formula the game logic uses. If a submitted score does not match the recomputation, the round is rejected. That is how every number on the leaderboard is made to stand up to checking, rather than to whoever edits the front end.

The full-mark line and the pass line for each tier are published on the level page, so everyone faces the same thresholds on every device. Prompts are generated server-side from a per-round token, and the server keeps a replayable record. In other words, a score reflects what you actually did in that round, not what you claim you did — editing the front end or resubmitting the same answers cannot buy you a better result.

The cognitive science, and its limits

These games come from classic tasks cognitive psychology has used for decades. Shape Memory and Colour Imprint sit close to short-term and visuospatial working-memory paradigms; N-Back is the standard working-memory updating task; Schulte Grid and Flash Click map onto visual search and reaction time; 24 Game, Sudoku and Sequence Reasoning train deduction and elimination. The tasks are used because their rules are simple and their results repeatable, which makes them stable self-test measures — not because they have been shown to make anyone smarter.

The limit needs stating plainly: this kind of training reliably improves performance on the same kind of task — near transfer — but there is no convincing evidence that it raises intelligence or transfers to everyday life. That is not modesty; it is where several large studies converge. So we do not promise that you will get smarter, that we prevent dementia, or anything medical of any kind.

Concretely: Owen et al.’s 2010 Nature trial had 11,430 people train for six weeks; they improved on every trained task but showed no transfer to untrained ones. The meta-analyses by Melby-Lervåg and Hulme (2013, 2016) likewise found only near transfer. Simons et al.’s 2016 consensus review found solid evidence for trained tasks but little for everyday cognition. And in 2016 the US Federal Trade Commission (FTC) ruled that Lumosity’s claims about improving school and work performance, slowing cognitive decline and preventing dementia were unsupported, settling for $2 million.

Privacy and data

We store only what the site needs to run: an optional account, your scores, and unlock progress calculated from the total number of days you have signed in. We do not collect information we have no use for, and we never ask for your real name or phone number. There is no reason for us to know who you are, and none of that data would help you play better.

Guests can play without signing up, at the cost of scores not being saved or ranked. The site carries no advertising and no third-party tracking scripts, and we do not profile or sell your data. Personal pages and the admin area are excluded from search engines, so your scores and details do not surface in public search results — that is deliberate, because results and personal details are private.

• Stored: account details, scores, unlock progress
• Not stored: real names, phone numbers, payment details
• Never: ads, third-party tracking, data resale
• Not indexed: personal and admin pages (noindex)

Content and updates

The text and difficulty figures on the level pages are generated from the real configuration in the code, not copied by hand. Difficulty tables read each game’s config directly, and score thresholds come from the same set of constants. Change the config and the page follows, so a page can never claim one thing while the game does another — and there is no hand-maintained copy to drift out of date.

Every page shows a last-updated date at the bottom, so you can judge how current it is. If the text disagrees with how a game actually behaves, or a number looks wrong, please tell us — we will check it, fix it and update the date along with the change. Mentioning the level, the tier and roughly when it happened helps us find the exact round.

Where to send it: the site’s public email is not set up yet, so please use a repository issue. A number anyone can independently verify is worth more to us than a sentence that sounds nice, so if we cannot support a claim we would rather change the wording than fudge it. That is also the only way we can keep these figures trustworthy over time.

What we deliberately don’t do

We do not promise that you will get smarter, that your intelligence will improve, or that dementia will be prevented or delayed — the previous section lists why. Any brain game sold on "getting smarter" is making a claim the evidence does not support, and we would rather not do that. If solid evidence ever appeared, we would change this sentence at once; until then, a promise would just be empty words.

We do not run "your brain age is X" assessments, which look precise but have no scientific basis. We do not use anxiety to sell things, either: no countdown pressure, no "act now or lose it" pop-ups, and no copy that tries to make you feel your scores are worthless. If a game is genuinely interesting, it should not need fear or urgency to keep people around.

Everything on the site is free, and we would rather not trade hype for traffic. We hope you treat it as a game you can keep challenging and watch yourself improve at, not a magic pill that changes your life. If a level stops being fun, tell us — making the games more replayable is worth far more than making the claims louder.