Plex Is the Cautionary Tale

Media servers are one of the classic homelab gateways. They start as a simple goal: put your own media on your own storage and watch it on your own devices. Then the details show up. Metadata, transcoding, remote access, clients, user management, hardware acceleration, offline viewing, mobile apps, and whether the server still feels like it belongs to you.

The three names that keep coming up are Plex, Emby, and Jellyfin. They all solve the same broad problem, but they do not represent the same philosophy anymore.

Plex: The Warning Label

Plex should have been the easy recommendation forever. It came from open-source roots, made home media servers approachable, and built the client ecosystem that everyone else had to compete against. For a while, Plex felt like the polished front end that self-hosted media needed.

Then it became the embodiment of enshittification.

The product drifted from “your media, your server” toward a bloated, account-centered, venture-capital-shaped platform trying to monetize attention around the edges. More hosted content, more upsell pressure, more features that serve Plex’s business model before they serve the person running the server. The experience became less about a clean self-hosted library and more about steering users through Plex’s ecosystem.

That is the part that makes Plex frustrating. The core product still has strengths. The clients are mature. The polish is real. But the direction is hard to ignore. Plex feels like a product trying to become a media company, an ad platform, and a subscription funnel, while the self-hosted server is treated as the legacy reason people showed up in the first place.

For a homelab, that is a bad trade.

Emby: The Better Plex

Emby is the clear better alternative to Plex for the way I want a media server to behave.

It feels more focused. It still understands that the server and the library are the point. The interface is familiar without feeling like it is trying to bury the local media experience under a pile of platform ambitions. It has mature clients, strong server features, hardware transcoding support, user controls, remote access options, and enough polish to be practical every day.

Emby is not perfectly open, and it is not trying to be. Some features require Premiere, and the project has its own commercial model. The difference is that Emby’s commercial model does not feel like it is trying to turn the product against the user. It feels like paying for a media server, not paying to avoid the consequences of someone else’s growth strategy.

That matters. A paid product is not automatically anti-user. A product becomes anti-user when the roadmap stops respecting what users came for.

Emby still feels like a media server first.

Jellyfin: The One To Watch

Jellyfin is the one with the most interesting long-term potential.

It is fully open source, community developed, and philosophically aligned with the original appeal of self-hosting. No premium unlocks, no vendor account requirement, no attempt to turn your server into a growth channel. It is the cleanest answer if the question is, “which one best represents the self-hosted ideal?”

The tradeoff is maturity. Jellyfin has come a long way, but it still needs time to bake. Client support, polish, edge-case handling, and some of the day-to-day smoothness still vary depending on platform and workflow. For some setups, it is already enough. For others, it is close but not quite there.

That is not a criticism of the direction. It is the normal cost of open development catching up to years of commercial polish.

Jellyfin is the new kid with the right instincts.

Where I Land

For my lab, Plex is the cautionary tale. It is what happens when a useful self-hosted tool starts chasing platform economics until the user-owned part feels like a feature inside someone else’s business plan.

Emby is the practical choice today. It gives me most of the polish people liked about Plex without the same sense that the product is trying to escape its original purpose.

Jellyfin is the one I want to keep watching. If it keeps improving, it has the best chance of becoming the default answer for people who care about open development and local control.

That is the pattern across the lab in general: pick tools that respect the operator. The best homelab software should make the system more yours, not less.