I feel the article is pretty clear that it is attempting to improve the BUSL.

1) It’s a license without any variable substitution (as the BUSL has).

2) It also has a firm change date to Apache 2.0 or MIT after 2 years (the BUSL default is 4, but this is also variable).

The BUSL has seen significant adoption but has a host of problems outlined in the article. But if you believe software should be an OSI-approved-license-or-nothing, this may not interest you.

Equating "Here's the full source code and you can do more or less what you want with it, except this one specific thing because it would really harm the future development of this software" with "proprietary" (that is, "no access to the source code at all") is just silly and ridiculous.

FSL has less ambiguity when compared to BUSL. Each use of BUSL is essentially its own unique license due to the variable Additional Use Grant. For example, a company could not pre-approve BUSL as an acceptable license for the company, because all BUSL licenses will have a different Acceptable Use Grant. On the contrary, FSL is a new license that can be pre-approved since it has no variables. It's an improvement upon BUSL. I personally like it, although I won't be moving from ELv2 to FSL due to another reason [0].

[0]: https://github.com/getsentry/loose-confederation/issues/4#is...