Argues that if you can't physically hold or fully control a piece of software, you don't truly own it. Every license-check endpoint, telemetry pipe, or cloud control plane is a unilateral kill switch the vendor can pull at any time, regardless of what you paid.
Frames the JetBrains, Docker, HashiCorp, Redis, Elastic, Unity, and MongoDB relicensing events as a single trend rather than coincidence. Argues the incentives — new CFOs, missed quarters, PE acquisitions — consistently point vendors toward extracting more from locked-in users, and this pattern will keep recurring.
Counters the common engineer reflex of 'I'll just self-host the OSS version' by pointing out that the relicensing wave specifically targets that escape route. BUSL, SSPL, and similar source-available licenses are designed to neutralize the self-host defense, leaving even technically capable teams with diminishing options.
Highlights JetBrains' 2024 decision to end perpetual fallback licenses on new subscriptions as emblematic of the shift. The mechanism that let subscribers retain a working version after canceling — a kind of soft ownership — is being systematically removed across the industry, leaving subscribers fully dependent on vendor goodwill.
A short essay titled *"If You Can't Hold It, You Don't Own It"* hit 388 points on Hacker News, and the comment thread — usually the place where HN goes to nitpick — turned into a wake. Nobody pushed back on the premise. Instead, hundreds of engineers cataloged the things they paid for and lost: Sony pulling Discovery content from PlayStation libraries, Ubisoft revoking *The Crew* from accounts that had bought it outright, Amazon quietly editing already-purchased Kindle books, Google killing Stadia and refunding everything except the hours.
That's the consumer side. The developer-side ledger is, if anything, longer. JetBrains ended perpetual fallback licenses on new subscriptions in 2024, meaning the IDE you've been paying for monthly stops working the day you stop paying — no permanent version, no grandfathered tier, nothing. Docker Desktop went paid for companies over 250 employees in 2021 with sixty days' notice. HashiCorp moved Terraform to the BUSL in 2023, prompting the OpenTofu fork. Redis followed in 2024. Elastic did it in 2021, un-did it in 2024, but the trust damage was already priced in. Unity tried retroactive per-install pricing in September 2023 and only walked it back after the entire indie ecosystem threatened to leave. MongoDB had already done SSPL in 2018. The list keeps growing because the incentives keep pointing the same direction.
The HN essay's framing — *if you can't hold it, you don't own it* — is older than software. What's new is the mechanism. Every modern developer tool ships with a license-check endpoint, a telemetry pipe, or a cloud control plane, and any one of those is a kill switch the vendor can pull unilaterally. You don't need to be malicious to use it. You just need a new CFO, a missed quarter, or a private-equity acquisition.
The defense most engineers reach for first — "I'll just self-host the open-source version" — is increasingly a fiction. The relicensing wave specifically targets the self-host escape hatch. BUSL, SSPL, the Elastic License, the Confluent Community License: every one of these is engineered to let the vendor say "open source" in marketing while forbidding the one thing that makes open source actually open, which is a competitor (or a sufficiently annoyed customer) running it themselves at scale. The forks exist — OpenTofu, Valkey, OpenSearch — but each one is a coordination problem that costs the community a year of velocity to solve.
The second-tier defense is "I'll pay, I just want stability." That worked when software was a product. It doesn't work when software is a contract. Sublime Text still sells a perpetual license that works offline forever; almost nobody else in the JetBrains/VSCode tier does, and that's not an accident — it's a business-model choice the rest of the market has rejected. When Atlassian killed server licenses for Jira and Confluence in 2021, customers who had paid for perpetual on-prem deployments got eighteen months to migrate to Data Center (at multiples of the price) or to Cloud (where their data lives on Atlassian's servers and their pricing changes annually). There was no "keep what you have" option.
The community's reaction in the HN thread is worth quoting verbatim, because the tone is the story. The top comments aren't outraged; they're resigned. Engineers are no longer treating relicensing as a betrayal — they're treating it as weather. That shift matters. Outrage forces vendors to walk things back (see Unity). Resignation just funds the next acquisition.
There's a practical procurement discipline hiding inside the philosophy essay, and it's worth writing down. Treat every tool in your stack as having a *revocation surface*: the size of the blast radius if the vendor changes terms tomorrow. A library you've vendored has near-zero surface. A self-hosted open-source binary you can fork has small surface. A SaaS where you can export your data has medium surface. A SaaS where your data lives in a proprietary format and the API is the export mechanism has large surface. A subscription IDE with no perpetual fallback has maximum surface — the moment payment stops, productivity stops.
The move is not to refuse SaaS or refuse subscriptions. The move is to know which bucket each tool sits in and make sure the high-surface ones are the ones you can actually leave. Concretely: any tool that holds your source code, your secrets, your customer data, or your build pipeline should have a documented export path that a junior engineer could execute in a day, and you should test it annually. If you can't, you don't own your stack — you're renting it under terms the landlord can change.
The second discipline is budget. Migration costs are real and almost never accounted for until they're urgent. The teams that survived the Terraform-to-OpenTofu transition smoothly were the ones who already had infrastructure-as-code abstractions thin enough to swap. The teams that survived the Docker Desktop license change were the ones who'd kept their dev environments reproducible without it. The pattern is the same: optionality is a line item, and the budget for it gets spent in calm weather, not during a fire.
The relicensing trend isn't reversing. The economics favor it: every VC-funded infrastructure company eventually needs to monetize the hyperscalers running their open-source code for free, and BUSL is the legally clean way to do it. Expect more of the canonical open-source infrastructure tools — observability, message queues, vector databases, the next generation of orchestrators — to ship under "source-available" licenses that look like open source until the day you become a competitive threat. The defensive posture for an engineering org in 2026 is to assume every dependency is on a clock, to prefer the truly-permissive over the source-available where the capability is comparable, and to keep migration plans warm. The kill switch is real. Knowing where it lives in your stack is the new ops hygiene.
Just pirate it. They can't tell you this but there's a quagmire of rights, licenses, agreements, treaties... and you can untangle this Goridan Knot by just pirating, especially media, for your own use.There are pixel perfect 4k drm-free rips out there made by people who poured thousands of
Since I don't see it mentioned yet in the comments:In 2011, movie studios created a digital ownership service called Ultraviolet. You could own titles in your "UltraViolet Digital Rights Locker" and access them from multiple devices via third-party streaming services. [1]"The Ult
Sony's one sentence notice is pretty grim considering how much money they made from these sales (sorry licensing).From September 1, 2026, due to our content licensing agreements, you will no longer be able to access your previously purchased content from Studio Canal, and it will be removed fro
Tangential, but a few days ago I started some Steam games I hadn't played in some years. I was surprised to be met with updated user agreements, which I had to agree to if I wanted to play the games I bought years ago. These were all single-player games.> If you can't hold it, you don&#
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I agree with the sentiment implied by the author, but I would reword it slightly. If you don't have the freedom to share something, you don't own it.I disagree with the interpretation that it needs to be held physically. Digital ownership is still ownership. I go out of my way to find musi