This week marks the commencement of some of the less pleasant work of organising a community conference - sending a decline email to many people who submitted a proposal to speak. Over the next few days we intend on communicating with everyone, regardless of the outcome of their proposal to the conference.
Talk declines are a necessary part of organising a conference. This year, we had more proposals than ever before, and that meant we also had to send more declines as well.
First, thank you to everyone who took the time to submit to the conference. The program review and selection process was incredibly difficult this year, and if you receive a decline email today please don’t take the rejection to heart. Some speakers (who submit regularly to conferences) tell us they celebrate a decline, as they can now enjoy the conference without the stress of presenting!
This post is an effort to be transparent about why talks get declined, and some common feedback we would like to share in the hope that it helps you submit improved proposals to community conferences - and to the next PyCon AU!
We published a post outlining our review process in a lot more detail, however here is a quick refresher of our process:
Through this process, the strongest proposals tend to ‘rise to the top’. While we do not select purely on highest mean score, it certainly helps talks get discussed earlier in the process before the schedule ‘fills up’. Additionally, every proposal was reviewed by at least 12 people this year, so no single review or score is enough to skew a talk to be selected.
It’s quite often the case that highly reviewed proposals don’t get consciously ‘declined’, it’s just that they didn’t quite get ‘accepted’ in time before the program fills up. It’s our privilege to have more great proposals than we can fit into our program.
There were several themes from reviewer comments that we want to share in aggregate. These themes were common across many proposals that received a decline email today.
Proposals that comprised only one or two sentences in each of the abstract and description were not likely to be reviewed favourably.
There’s no specific rule of thumb for word count or length. Our reviewers are looking for enough detail that they understand how you’ll fill a 25-30 minute time slot with an interesting, informative and engaging talk.
Some choice comments from our reviewers include:
An awesome talk title with excellent puns or alliteration is not enough to carry a proposal through to being accepted. Please tell us what you’re actually going to say!
Some proposals gave us a LOT of detail. Too much so. It was common in these proposals to see extensive use of headings and other markdown formatting. In a similar vein, there were a large number of proposals that had similar formatting that were very likely full LLM output copy and pasted.
Too much detail leaves the reviewers asking two real questions:
While PyCon AU only offered 30-minute sessions this year, we have previously offered longer 50 minute deep dives that might be more appropriate for proposals of this length. Some proposals acknowledged their own uncertainty about the time allocation, but did not iterate on the proposal enough to address that concern before the CFP closed.
Some proposals were well structured but didn’t outline a clear set of take-home insights, skills or talk narrative. Attendees look at the abstract and description and make a decision on which of three talks to attend at any given time so the topic, take home skills and main thesis of the talk need to be clear and succinct. The reviewers look for the same information too.
The comments in these scenarios were a little more varied, because the reason a talk outline is unclear is (ironically) also not always clear.
Rather than submit a single refined proposal, some speakers submitted several variations on a theme. While we did not set a limit on the number of talks submitted, we did see a trend where multiple proposals from a single speaker often led to proposals being reviewed as ‘more average’ overall.
We get it, it’s easy to worry that you didn’t get the phrasing or the topic quite right, and maybe this slightly different angle will be better. Our working thesis as a conference is that the scatter gun approach to submitting multiple similar proposals doesn’t encourage or allow speakers to fully consider and refine each of their distinct talk ideas to the best submission it can be.
If you’ve got a great idea for a topic and you aren’t sure how to submit then please get in touch with one of our mentors to discuss, or attend a CFP workshop to refine the idea before CFP closes.
Many proposals this year self-identified that they used AI to write their proposal. (Many more clearly did, but also went to the effort of editing and refining before submitting).
We absolutely acknowledge that leveraging AI tools is one of many techniques speakers use to break through writers’ block and brainstorm privately their talk ideas, but we equally value proposals that are thoughtful, personalised and self-aware.
I’ll quote just one comment from the review process that stood out to me as director: “if you can’t even be bothered to mask the fact the AI wrote this for you…”
We intend on providing clearer guidance to accepted speakers on the conference’s policy regarding thoughtful and transparent use of AI when presenting at PyCon AU 2026.
To quote our sponsorship prospectus: “Attendees at PyCon AU expect high quality educational content, not just a product demo“.
Many talks were submitted by employees on behalf of for-profit companies, with somewhat transparent (and often blatant) product placement. To be clear, we do not blindly decline talks mentioning products:, in fact several highly reviewed proposals were discussed in depth by the review committee.
The decision to accept or decline a proposal in this category stemmed on two key questions:
Where it gets murky is when companies also run, publish or oversee projects and packages that are ‘open source’ or have various free tiers. This post won’t get into the nuance, technicalities or ethos of ‘open source’, ‘code available’, ‘free’ and ’open’. I mention it here to acknowledge that we are hyper aware of both the bridge between and the murkiness underneath these types of projects (and the talks proposed about them).
While it is difficult to provide example reviews without citing specific companies or products, proposals that mention a paid product as an afterthought or reference case study at the end were more likely to be considered, especially if the proposal was thoughtful to consider alternatives for evaluation.
We reserve a handful of scheduled talks for sponsors who contribute financially to PyCon AU. Product placement, demos and case studies are best suited to companies who support the conference via sponsorship.
As an all-volunteer team, we are cautious to commit to every speaker to provide feedback on every proposal we had to decline. However we also acknowledge the important value that feedback has on encouraging first time speakers and supporting our community to improve the overall quality of our future conference program.
If you’re reflecting on a declined proposal and the above notes don’t quite land as relevant to your submission, we are happy for speakers to ask for feedback on one of their submitted proposals.
To ask for feedback, please fill in this form. Please use the same email you did for Pretalx. You will need your proposal ID which will be included in the decline email. (It’s also in the URL of your submission when viewed in Pretalx).
If you submitted multiple proposals that were declined, think about which proposal you would most like to improve on and submit the form for that one proposal.
It may take us quite some time to work through all submissions - potentially several weeks. We are, after all, still volunteers working to run a conference in August. Thank you in advance for your patience while we work through your requests for feedback.
Please note: individual scores and review comments are considered confidential and we won’t share reviewer comments verbatim. Instead, we will take time to reflect across the reviewer comments and share balanced feedback in the hope that it is more actionable.
When we opened CFP we had hoped to finalise the logistics and costs of running a poster session by the time review had closed. Unfortunately, various circumstances have led to us needing a little more time to finalise our planning for posters. Rather than delay - we’re also emailing everyone who submitted a poster with an update.
The silver lining to this delay is we get a little more time to consider new proposals that might work better as a conversation and a poster than a talk.
If you think your proposal would work well as a poster at the conference, please consider submitting an expression of interest with your talk’s proposal code.
Please note: the CFP FAQ and guidelines for posters and benefits still apply, and we are unable to offer a complimentary ticket to the conference for poster sessions.