There was a time when a game shipped and that was that. What was on the disc was the product. Patches barely existed and the studio moved on.
That model is gone, and the change has been more thorough than in any comparable medium. A game today is frequently a service that begins at release rather than concluding there.
What early access actually solved
Worth starting with the case for it, which is stronger than the complaints suggest.
Game development is expensive and long. A small studio building something ambitious faces years without revenue, which historically meant either publisher funding — with the creative control that entails — or not making it.
Early access lets a studio fund development from players who want the thing to exist. That has directly enabled a category of game that the traditional model wouldn't have produced: mechanically complex, systems-driven, niche.
The other genuine benefit is iteration. Some designs can't be evaluated without a large player base doing unexpected things. Balance in a complex simulation, emergent behaviour in a multiplayer system — you can't test that with twelve people in an office.
Several of the most acclaimed games of the last decade were built this way, in public, over years, shaped substantially by player feedback. That's a real achievement of the model.
What it broke
The obvious problem: "finished" stopped meaning anything.
A game leaves early access when the developer says so. There's no external standard. Some titles emerge polished and complete; others leave early access in a state that would have been considered unshippable fifteen years ago, and continue being fixed afterwards.
The consequence for buyers is that release date carries no information about condition. You have to research the specific title's state, which is a burden the old model didn't impose.
Then there's abandonment. Games enter early access, take money, and stop being developed. Sometimes the studio fails, sometimes interest wanes. Buyers are left with a permanently unfinished product and generally no recourse beyond a refund window that expired long ago.
The full-price version of the same thing
The genuinely objectionable variant isn't early access at all. It's full-price releases from large publishers that are functionally in early access without saying so.
A major title released broken, with a roadmap of fixes, at a premium price, is doing exactly what an early access game does while presenting itself as finished. That's happened repeatedly and it's the practice that deserves the criticism early access as a category tends to attract.
The difference is honesty about status. An indie studio selling an unfinished game and labelling it unfinished is being straightforward. A large publisher doing the same thing without the label is not.
Day-one patches and what they conceal
A related mechanism. Physical media has to be manufactured weeks before release, so the disc contains an older build and a patch downloads on launch day.
Sometimes that patch is a modest set of fixes. Sometimes it's larger than the game on the disc, which means the physical copy is close to a licence key.
Practically this matters for preservation. A disc that requires a patch to function is dependent on servers that will eventually be switched off. A generation of games may become unplayable in their released form once distribution infrastructure retires, and it's a genuine issue that archivists have been raising for years without much response.
What good early access looks like
Having watched a lot of these, the ones that work share features.
A clear, honest statement of current state and what's missing. Not marketing copy — an actual list.
Regular, visible development. A public changelog and a pattern of updates. Silence for six months is the strongest predictor of abandonment.
A core loop that's already fun. Early access should be about expanding and refining something that works, not about finding out whether the fundamental idea does. Games that enter without a working core rarely find one.
And a realistic scope. The projects that fail are usually the ones promising far more than the team could plausibly build.
How I approach it
Simple rule: buy an early access game if the current state, exactly as it is, is worth the price to you. Treat every promised feature as unlikely to arrive.
If it does arrive, that's a bonus. If it doesn't, you paid for something you were happy with.
That framing removes most of the disappointment from the model, and it also means you're supporting developers whose current work is already good — which is probably the selection pressure the whole system needs.
The refund window problem
One structural detail deserves more attention than it gets: refund policies are typically based on playtime and elapsed days since purchase. Both are short.
That is a reasonable rule for a finished product. For an early access title bought on the strength of a development roadmap, it is close to meaningless — the promises being evaluated extend over years, and the window to act on their failure is measured in weeks.
Effectively, buyers are asked to assess a multi-year commitment inside a fortnight, and after that the risk is entirely theirs. Nobody has proposed a good fix, and the mismatch is probably the single most consumer-unfriendly feature of the whole model.