Hi Jie, Guangdong, and Twice, Thanks for the detailed responses. I am keeping my -1 (binding), and I would like to explain the reasoning more precisely and respond to the points raised.
Twice, one small note first: when referring to concerns already raised on this thread, it would be more helpful to quote or link the original message rather than write "some people are concerned". My -1 is on the record with specific commit links, and summarising it as an anonymous concern makes it harder for others to check what was actually said. Not a big deal, just easier for everyone to follow. First, On the project history Jie's reply clarifies several things: the dense commit sequence between May 20 and May 23 comes from a single-developer bootstrap phase; development was heavily AI-assisted; some Git author identities refer to the same person and the raw author count needs normalisation; and development and review activity is still concentrated among the initial contributors. I appreciate the transparency here, and it resolves my question about whether activity had been inflated. It also settles the factual picture: as of today, Maka has a codebase and a small number of developers, and the community described in the proposal is still mostly forward-looking rather than existing. That is the basis for my vote, not the AI tooling itself. Second, On deferring provenance to IP clearance Guangdong mentioned that provenance will be reviewed as part of the code transfer and IP clearance process. I agree that process will run, but I don't think it covers the specific question here. SGA and ICLA establish that a contributor has the right to license what they submitted. They do not establish whether a large, heavily AI-assisted initial import contains material that is incompatible with the Apache License. Under the ASF guidance on generative tooling (https://www.apache.org/legal/generative-tooling.html), that responsibility sits with the contributor, and it is easier to address before the code enters the ASF than after. So I would like to ask, ideally during the vote window: 1. Roughly what share of the initial import was AI-generated or AI-assisted, and with which tools? 2. What review was applied to those outputs with respect to license compatibility, and is that review recorded anywhere reviewable? 3. If provenance cannot be established for some parts, is the plan to remove and rewrite them, or to keep them pending discussion? If the answer is that all of this will be handled during incubation, then the practical effect is that the mentors and the IPMC absorb the cleanup work for a codebase whose provenance is still open. I would rather we know the scope of that work before the vote closes than discover it afterwards. Third, On "this is what the Incubator is for" Twice, I understand your argument, and I agree the questions you raise are worth answering. I read the Incubator's role a bit differently: it exists to help an existing community learn to operate the Apache Way and to normalise IP and governance, rather than to test whether a community can be grown from a standing start. When the latter does not work out, the cost falls mainly on mentor time, IPMC review capacity and the ASF's name. You also noted that the Incubator has little experience with this kind of project, that it may challenge some existing ASF assumptions, and that additional graduation metrics might be needed. I think those are policy questions, and they are better settled on general@ as policy than settled implicitly by accepting one proposal. My suggestion would be to discuss what we expect from AI-native projects first, and then evaluate proposals against that. If the IPMC does decide to accept Maka as a first case of this kind, I would ask that the following be written into the proposal before acceptance rather than left to later interpretation: - A concrete independence target, e.g. active committers from N mutually independent organisations before graduation. - A time box and exit condition, e.g. if the target is not met within 12 months, a retirement discussion is opened. - A disclosure and review requirement for AI-generated content. - A requirement that technical decisions are made on the mailing list; GitHub PR discussion alone does not satisfy this. With those in the proposal, I would be much less concerned about the precedent. The last, What I would suggest instead: I have no objection to the technical direction, and the problem area Guangdong describes is real. My concern is timing. A safer path would be another 6 to 12 months in the current repository, turning the following from plans into facts: provenance of the AI-assisted portions documented, two or three sustained contributors from different organisations with real review weight, one or two public releases, and design discussion moved to a public list. A proposal after that would answer nearly every objection raised in this thread on its own. I understand acceptance votes are decided by majority and that my -1 does not block anything, and I am not trying to hold up the process. My aim is to have these points on the record and tracked explicitly in the mentors' and PPMC's early reports. On Mon, Aug 10, 2026 at 4:09 PM Twice <[email protected]> wrote: > > I saw that some people are concerned about the project's heavy use of > AI in development, and whether this model can really build a genuine > community. > > But I think this is exactly why the ASF Incubator is meaningful. We do > not really know whether this development model can be sustainable in > the long term, or whether it challenges some of ASF's existing ideas > and principles. That is precisely why it is worth exploring. > > On one hand, ASF seems to have few general-purpose agent projects > today, so bringing in something new in this area may have value and is > worth experimenting with. > > On the other hand, the Incubator itself also has little experience > with this kind of project. How should such a project develop during > incubation? Should there be any additional metrics or requirements for > graduation? If we can gain more experience with these questions > through incubation, I believe that would be useful for ASF. > > So, +1 from me. > > Also, could you add me to the mentor list? I am quite interested in > this project for the reasons above. > > Best, > Twice > > On Mon, Aug 10, 2026 at 2:33 PM Guangdong Liu <[email protected]> wrote: > > > > +1 (binding) > > > > I have reviewed the Maka proposal and the discussion. I support > > accepting Maka into the Apache Incubator. > > > > Maka has a coherent technical direction: a local-first and > > provider-neutral agent runtime and workspace, with Desktop, CLI/TUI, > > and headless surfaces built around durable execution records, explicit > > permission boundaries, and recoverable task and session state. This is > > a meaningful problem area with value beyond any single model provider > > or hosted control plane. > > > > The project also brings an existing public Apache-2.0 codebase and > > public project materials for contribution, security, architecture, > > design, changelog, and notices. I believe it has sufficient technical > > substance and an initial community to begin the Apache incubation > > process. > > > > I recognize the concerns raised about the project's young history, the > > early contribution and import pattern, code provenance, and the need to > > demonstrate a broader and more independent contributor community. > > These are material concerns. I do not view incubation as evidence that > > they are already resolved; rather, I expect them to be concrete early > > milestones for the PPMC and mentors: public and verifiable provenance > > and IP work, mailing-list-based decision making, transparent governance, > > and sustained growth of independent contributors and committers. > > > > On balance, I believe Maka is suitable to begin incubation and that the > > Apache Incubator provides the right framework to address these matters > > in public. > > > > Best regards, > > Guangdong Liu > > > > On 2026/08/09 18:20:21 tison wrote: > > > Hi everyone, > > > > > > Following the discussion thread [1], I call > > > for a formal vote to accept Maka into the Apache Incubator. > > > > > > As a reminder, the Maka Proposal can be found at [2]. > > > > > > [1] https://lists.apache.org/thread/0n79l7dbz9jm694rpnbrxtn8f6lmnw1j > > > [2] > > > https://cwiki.apache.org/confluence/spaces/INCUBATOR/pages/446070971/Maka+Proposal > > > > > > Please cast your vote: > > > > > > [ ] +1, Accept Maka into the Apache Incubator > > > [ ] 0, I don't care > > > [ ] -1, Do not accept Maka into the Apache Incubator, because ... > > > > > > The vote will remain open for at least 72 hours, and only votes from the > > > Incubator PMC members are binding. Votes from everyone are welcome. > > > > > > Thanks! > > > > > > Best, > > > tison. > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [email protected] > > For additional commands, e-mail: [email protected] > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
