On Tue, Oct 13, 2026 at 01:49:36PM +0100, Lorenzo Stoakes (ARM) wrote: > On Tue, Oct 13, 2026 at 01:19:40PM +0200, Jori Koolstra wrote: > > > The only long-term solution to that is to hit mark read 1,000 times :) > > > > There's a way to nuke drive-by authors and that is to require them to do > > some kind of "proof-of-effort." So you want to contribute to the kernel? > > Good, come to one of the OSS conferences, or LPC, or have some well-known > > maintainer vouch for you, or do the kernel mentorship program. The point > > is not that this cannot be gamed, but that the "gaming" in itself requires > > some kind of investment, like proof-of-work. > > Yeah that is a neat idea :) like universities requiring people to be in > person for exams and shifting grading weight mostly to that.
That's a bad analogy. Try telling those university students that will have to travel across the globe to sit that exam, and pay for it from their own pocket. > Let's not encourage people to ask maintainers for backing though! Maybe > leave that bit unsaid. > > > We may miss the savant that for whatever reason cannot show up to a > > conference, > > but Linux Foundation is already very supportive of funding people. And > > besides, > > Linus has already made it clear that we can expect of people to be able to > > grow > > into serious community members (which may include being able to effort time > > and/or > > money for LLM access.) > > True. I did worry about people from countries/with lack of means where they > might struggle with this requirement, but LF help I guess addresses that. I'm willing to discuss that only based on hard data. > > This would solve most slop immediately, and we will know about LLM found > > bugs > > anyway from cooperation with Anthropic and just the maintainers using LLMs. > > In mm we're more and more dealing with bug 'fixes' by taking over the > patch, which is its own workload but will figure out how to do TODOs for > that in time. -- Regards, Laurent Pinchart
