Own the Library, Rent the Player

Nothing broke. The server is running. The drives are spinning. Every file you ripped, tagged, and organized over a decade is still sitting exactly where you put it. The software just quietly stopped putting you first.

In July 2026, Plex moved its Lifetime Pass from $119.99 to $749.99 and introduced a five-year tier at $249.99. Anyone who bought in at $75 during a 2021 sale is now sitting on what looks like the best deal they ever made. That feeling — I got in early, I'm safe — is worth examining. A lifetime pass is a guarantee about the price, not about the product. It locks in what you pay. It locks in nothing about what you get.

---

i · drift isn't betrayal. it's budget.

The instinct is to read this as a story about a company turning greedy. It's more useful to read it as physics.

A platform has to fund itself. As it matures, the cheap growth is gone and the expensive obligations remain — apps on nine TV operating systems, authentication servers, transcoding infrastructure, support. The pressure to convert users into revenue doesn't come from a villain in a boardroom. It comes from a spreadsheet that has to balance. And the spreadsheet reshapes the product: the home screen surfaces ad-supported channels before your library, hardware transcoding sits behind a subscription, remote streaming gets its own pass, features with no revenue attached quietly get deprecated.

Expecting a company to permanently resist that pressure is a category error. It's moralizing at a system instead of designing around it. The useful question was never will they stay good? It's what happens to me when they change?

That's a design question. Which means you can answer it before you need the answer.

---

ii · the exit was always there. that's not the same as being able to take it.

Jellyfin has existed since 2018. Free, open source, local authentication, no tracking, hardware transcoding included, no ad row on your home screen. It has been sitting there the entire time as a fully-formed alternative.

So why didn't people leave?

Because exit is a capability, not a decision. On the day the price hike lands, you're angry, and anger is not a currency you can spend on migration. What migration actually costs is: an evening of learning a new server, a weekend of re-verifying that your file naming parses correctly, a hard conversation with three family members about a new app on their TV, the loss of eight years of watch history, and the discovery that the client on your specific 2019 smart TV is a volunteer project with fourteen open issues.

None of that is payable on the day you decide. It's payable in advance, or not at all.

This is the shape of the whole pattern: the escape hatch was always there, and taking it costs work — work that comes due at the exact moment you're least able to do it, when the anger arrives and the switching costs have compounded to their maximum. Sovereignty isn't a decision you make once. It's a capacity you maintain.

---

iii · the lifeboat has a leak too

Here's the part that keeps this from being a smug open-source lecture.

In the same month Plex raised its price, Jellyfin lost three of its most senior people. Project leader Joshua Boniface stepped down effective July 19, 2026, citing burnout: "I simply could no longer provide the effort (mental or time-wise) that the role demanded." Core team member Anthony Lavado — who handled app store operations, releases, and finances — left the same day. Co-founder Andrew Rabert resigned two days earlier, after friction over his development approach on a new desktop client.

The causes are worth getting right, because the obvious story isn't quite the true one. Back in May, the team said publicly that "increased support requests, combined with the AI code submissions, have led to burnout at various levels of the development and admin team" — machine-generated pull requests arriving faster than volunteers could read them. But Boniface, writing his own account, was careful: the arguments over AI were "not the sole, or even the main, reason" he left, just "another straw on the proverbial camel's back." What actually wore him down was two years of accumulation — being a reluctant people manager, saying no, writing policy, taking criticism personally. And Rabert's exit ran the other direction entirely: the friction was over his own use of AI-assisted tools.

So the same technology showed up on both sides of one small project — as unreviewed volume dumped on reviewers, and as a tool a co-founder was pushed out for using. That isn't a clean parable. It's what a load-bearing structure looks like when the load is real and nobody is being paid to carry it.

The project continues. A deep bench of contributors remains, and Boniface describes the transition as amicable. But the lesson stands, and it's the opposite of comfortable: there is no permanent safe harbor. The corporate platform drifts because it has to fund itself. The volunteer platform wobbles because it doesn't. Both are structures with failure modes; they just fail differently.

If your plan is "pick the trustworthy vendor," you have not made a plan. You've picked a horse.

---

iv · own the layer: your library is a directory tree

Here's the move, and it's smaller than it sounds.

Plex is not your library. Jellyfin is not your library. Your library is a directory tree on a disk, and both of those programs are just readers pointed at it. Own the layer that holds your data and intent; rent the thing that renders it.

That layer becomes portable the moment your files follow the conventions both scanners already understand:

Movies/
  Arrival (2016)/
    Arrival (2016).mkv
TV/
  Severance/
    Season 01/
      Severance - S01E01.mkv

That's the whole interoperability standard. Get your tree into that shape and any competent media server — Plex, Jellyfin, Kodi, whatever replaces all three in 2031 — will index it without being asked twice. Get it wrong and every migration becomes a bulk-rename project you'll postpone forever.

Push as much metadata as you can into the layer rather than the app. Artwork as poster.jpg and fanart.jpg next to the video file. Descriptions and cast in .nfo sidecars. Jellyfin reads NFO natively; Plex needs its local-media-assets agent enabled, or a scanner plugin. It's slightly more work up front. In exchange, your metadata survives the death of any particular server.

---

v · know exactly what doesn't port

Being honest about what stays behind is more useful than pretending nothing does.

Portable (lives in your tree): files, folder structure, local artwork, NFO metadata, subtitle sidecars.

Not portable (lives in the platform's database): watch state, resume positions, play counts, ratings, playlists, hand-curated collections, and user accounts.

That second list is the real switching cost, and it's the one people underestimate. Ten years of "what have I already seen" is genuinely valuable and genuinely trapped.

There's one more thing that doesn't port, and it isn't in a database at all: remote access. Plex brokers the connection for you — you sign in, and for most people it simply works. Jellyfin hands you the problem: a port forward, a reverse proxy, or a private tailnet of your own. These are solved problems with good options, and the tailnet route is genuinely better once it's running. But it is an evening of work, and it is the single most likely entry on your eventual list of things that broke. Discover it on a quiet Tuesday, not on the day you're angry.

For watch history, two things help. First, know the tool's name now, while you're calm, rather than searching for it mid-migration: WatchState (arabcoders/watchstate) is self-hosted, actively maintained, and syncs play state across Plex, Jellyfin, and Emby without routing your history through anyone else's server. Alternatives exist — JellyPlex-Watched, Jellex — though both have been quieter lately. All of them are volunteer projects. None is official. That's the honest state of it, and it's better to know that in advance than to assume a safety net you never checked.

Second, and better: don't migrate at all.

---

vi · run both. then run the drill.

The strongest position isn't Plex or Jellyfin. It's both, pointed at the same read-only mount, indefinitely.

The cost is a container and some RAM. Mount the tree read-only for at least one of them — a scanner with write access can rename files out from under the other server. The benefit is that the migration is already done. There is no switching day, because both systems are always current. When one drifts, you change which icon you tap. Nothing else moves.

But installing it isn't enough. An exit you've never used isn't an exit — it's a bookmark.

So run the drill. One evening, on purpose:

1. Point the second server at the same library tree.
2. Install its client on the single device that matters most — the living room TV, not your laptop.
3. Watch one thing end to end. From outside the house, if you stream from outside the house.
4. Write down every single thing that broke or annoyed you.

That list is your actual switching cost. Before the drill it was a vague dread; after the drill it's four bullet points, two of which you can fix in twenty minutes. Fear becomes a work queue. That conversion is the entire point of the exercise, and it's the same reason we test backups instead of merely having them.

---

vii · fund the escape hatch

One last thing, and it follows directly from the Jellyfin news.

If an open-source project is your insurance policy, an unfunded project is a lapsing policy. The people keeping your exit viable are absorbing support load and review load from users who give them nothing at all. Five dollars a month to the project that keeps your exit open is not charity. It's a premium on your own sovereignty, and it's the cheapest line item in this entire setup.

The review load deserves its own note, because it's the one you might be adding to. Opening a pull request full of unread machine output isn't contribution; it's cost transfer wearing the costume of help. And notice the shape of it — the same tool, in the hands of someone who actually read the diff, is just a tool. Amplification is only a gift when someone tuned it.

---

The platforms will keep drifting. That's not a moral failure, it's what happens when funding pressure meets a maturing product — and it will happen to whatever you migrate to next, in a different direction, for different reasons.

You can't buy a permanent home. You can only keep a standing exit: files in a shape anyone can read, metadata that lives beside the media instead of inside the app, a second server already running, a drill you've actually rehearsed, and a few dollars a month to the people maintaining the door.

Sovereignty was never a purchase. It's a maintenance practice — and the good news is that it's a small one, if you start before you're angry.