We're here!
Polkadot has scheduled it's first two batches of parachain auctions, and Kusama has just ended it's second batch.
So as of right now, the plan for Kusama's parachain slots and auctions are uncertain.
**tl;dr: the reasons we were manually executing and batching auctions no longer apply, and we should run scheduled and continuous auctions going forward. I propose a continuous weekly auction that can be paused in the case of extreme circumstances. **
During discussions over the last few weeks, I think we have established that the original reason to batch parachain auctions - to monitor for instability after each batch - is no longer a pressing concern. We've also established the feasibility (albeit, with some hiccups) of scheduling auctions for the future. Polkadot is even using the scheduler to run their first two batches of auctions, albeit they are restricted to two sets of five auctions, instead of a single continuous schedule.
I believe that, in the same spirit as Kusama's ramp up to 1000 validators before Polkadot, we should be ramping up to 100 parachains as well, ahead of Polkadot.
There are only 48 full weeks in the longest possible parachain lease period on Kusama, so even if we had a full schedule of an auction a week continuously, forever, we'd end up with a maximum of 48 simultaneously operating, auction-assigned parachains. I see that as both a reasonable soft cap on the number of auction-assigned parachains, pending changes to the logic or rate of auctions, and a reasonable rate at which auctions should run - seeing as Polkadot seems to be implying a cadence of an auction once every two weeks (which would, if they kept up the pace continuously, lead to the same number of simultaneously operating auction-assigned parachains). Under those conditions, if we want to more properly accomplish the purpose of being a Canary for Polkadot, I'd probably even suggest shorter auctions happening more often, so that we remain ahead of Polkadot and pushing the limits of the tech - but I don't believe there has been any suggestion that Polkadot will be operating continuous auctions as of yet, so there likely will be some leeway (there is a ratio of 75% of Kusama parachain slots to Polkadot parachain slots recommended in Parity's post on the topic of the Polkadot auctions).
All that said, I'd like to hear opinions on proceeding ahead with continuous auctions.
Here are some concerns we've already heard:
We should not change the precedent of 5 batched auctions and a break because it subverts people's expectations.
We should not change the current cadence of auctions on Kusama because it will change how the incentives for auction participants play out.
Continuous auctions may mean that it will be harder for contributors to predict how long their contributions will be locked.
Continuous auctions will have less hype/fund-gathering time between them, potentially leading to "cheap" parachain slots.
Continuous auctions may lead to crowdloan contributors locking up their funds in earlier auctions, leading to them being unable to participate in later ones.
Continuous auctions might lead to uncertainty from parachain teams regarding which lease slots to target, and how to configure their crowdloans.
The counterpoints to which are:
Expect chaos :) but more seriously, the initial parameters were fairly arbitrary - and we should be exploring, not sticking to an arbitrary initial precedent.
Same goes for game theory and incentives, we should perhaps be shifting towards long term planning and games rather than the short term sprints that batches represent. Participants should be supporting projects they want to see succeed, not guessing which project is likely to succeed and contributing to ride it's coattails - I would argue the race-like nature of batches prefer the latter, and continuous auctions need more planning on the part of teams and contributors.
Slot durations and contribution lockup times are transparent and predictable, it will require slightly more communication effort on the side of parachain teams however, because these details will likely no longer be uniform.
Cheap is good - we want to both grow the network and serve the purpose of somewhere with lower barriers to entry than Polkadot - that wouldn't be possible without an increase supply of parachain slots. Additionally, if parachain slots are highly in demand, then bids and slot bonds will reflect that with or without batching.
As stated earlier, contributors should plan carefully and support the projects they want to see successful, not the projects that come into the process likely to succeed. It should not be a problem to miss out on a handful of second best choices if your favorite project wins a slot, or in other words, just like the incentives around validating and council voting work out, it should be easier for a dedicated but small community to secure space/a voice on the network than an apathetic but large community.
Parachain parameterization is something the current model allows very little flexibility in :) we see everyone aiming for the longest possible lease period, starting in the current lease period, aiming for fast contributions, which benefits a certain marketing style and business model.
I'd like to see projects with longer term, but potentially still profit oriented goals, aim for later slots via auction (and projects with no profit incentive, should be applying to governance for common good slots). Or projects with more ephemeral goals aiming for a single lease period to launch their project, do their work, and spin down - issue specific temporary chain-DAOs galore. (As an aside: It would be amazing to see a hyper-decentralized DAO project that launches by spinning up a parachain for a more easily acquired single lease period, funded by standard crowdloan, which as a parachain, sets up and launches a trustless non-standard crowdloan hosted with custom terms and logic on their own chain.) Things like that are much riskier to attempt to run without a regular and continuous auction schedule, as crowdloan parameters can be left invalid and participants left hanging if there is no ongoing auction for the crowdloan pallet to bid on.
[Sorry, that was alot of text in order to say that continuous auctions are actually easier for teams to determine parameters for than the status quo.]
Instead of / in addition to the proposal for a weekly auction, we can probably also consider:
Things that are probably technically unfeasible at the immediate moment:
Other thoughts and alternatives to a weekly auction schedule to stay ahead of Polkadot are welcome.
The proposal for ramping up over slightly over 48 weeks to 48 parachains was made as one of the least extreme and disruptive possibilities.