Hacker News Reader

Comment by Seirdy | Hacker News Reader

Software that isn't executed on a user's machine is out of the user's control. The GNU AGPL redefines distribution to include serving over a network so that software served over a network must disclose source code as well (unminified JS, server-side code, etc). It's also copyleft, which is good: the purpose of permissive licenses is to enable proprietary derivatives, which is against the spirit of the Fediverse.

View on HN

One interesting consequence of Mastodon's AGPL licensing is that Gab, the right-wing twitter clone that runs a Mastodon fork, also has to disclose its source. That was how people found a vulnerability in Gab that allowed the recent "hack".

As a service operator, why would I want to choose software licensed this way?

As an operator you're always making a tradeoff between the "negatives" of a license and the "positives" enabled by the actual software.

So for one example, the recent (bait and) switch of Elasticsearch to SSPL effectively means that for many actual real-world users of Elasticsearch, the benefits of staying on the official Elastic branch has ceased to be worth the costs of the license when your alternative is running FOSS Elasticsearch/Opensearch.

So to answer your actual question - presumably as an operator you want to use the AGPL-licensed code more than you want to avoid those same negatives of AGPL.

That's exactly why I license everything I open-source MIT/Apache.

Because what you're offering is a service, and not someone else's software. Your added value is not actually in the code or the cute pictures; it's being available, having a low latency, allowing people with no tech knowledge to take part in the discussion, building a relation based on mutual trust rathet than serfdom. It's fundamentally the difference between a user and a consumer.

Reply on news.ycombinator.com