It was proposed in May and with initial budget of the CoE with not much development. I think it would have much bigger impact now than before with recent developments of a marketplace (NumerBay), Kashi Lending, more text/video content from the community to get started, and an increased funding from Numerai. Leading to more passive participation.
Since I can not edit the previous proposal, I’d like to make a new proposal with clearer goals and impacts that’ll benefit everyone.
Proposal
To make an incentive based task on 1729.com for user engagement. I have already drafted the task (explaining Numerai and task details). It rewards top 20 submissions with $100 NMR based on CORR +2xMMC. the best scores in 4 rounds after it’s published(for eg, the best CORR + 2xMMC in any of 270, 271, 272, 273. Thus, it lasts for 2 months to resolve). Thus, $2000 worth NMR in total. The goal is to release the task on Monday so new users have a week for experimentation before making a submission. However, I do not have any information on when it’ll be released if it’s accepted.
since 1729.com has higher reach to a broader audience working on Blockchain and Machine Learning, this task would put Numerai to a bigger appropriate audience and hopefully will help in Token appreciation by a larger audience.
With market place and Kashi Lending, more people can participate passively by providing liquidity (This is good for users) and by buying and selling predictions (that’s helpful for the meta-model). New users would have video/text content to get started.
Payout factor is also decreasing, so we have this window to engage more users and grow the community before it converges to 0.2 .
Timeline
The draft is ready which is just explanation of Numerai and the task details. If the task is released on Monday, the submissions will be made next Saturday. The round takes 4 weeks to close. So, we’d have winners after 5 weeks.
Best case outcome
Many new users join. Top 20 rewarded $100 in NMR (will be distributed by 1729.com , preferably on Coinbase wallet with instant transfer). Hopefully, some of that will stay in the tournament through stakes.
Word of mouth sharing from some influential Crypto and ML twitter handles.
Kashi Lending increases.
More buyers on the marketplace.
More viewers on content.
Worst case outcome
No one submits (highly unlikely)
Only 20 submissions, and all rewards are sold. No user turnout.
Success criteria
If the team has some kind of analytics to see how many sign-ups are coming form the task post, we can see how many users joined and set a baseline for counts.
Funding required
Yes, 1729.com tasks needs at least $1000 in Coinbase supported cryptocurrencies. I am proposing $2000 for this task.
I was working on this before the CoE was formed. As I am one of the CoE, I do not want any personal compensation.
Revenue
There is no personal monetary benefit tied to this. If this works, everyone equally benefits from this. If it doesn’t, no one loses anything (except the funding and my hosting).
If this works, we can have tasks for Signals with bigger funding.
Note:
Currently, as I am the proposer and a part of CoE, I will not vote on this.
I’ll not initiate/queue up the transaction.
Other elders have to start this and fund it.
Funds will not go to my wallet, instead directly transferred to 1729.
This, according to me is a very high RoI experiment.
Is there a way to give people more time than just one round? Having >>20 people see the competition, learn what Numerai is and how it works, then submit something unique and useful within 5 days seems very very tight. Maybe best round in a 4-round window? I see the potential benefits you outlined and my worry would be that people would see the very short deadline for the complicated task, think it’s futile and move on.
Would the website accept crypto for our funding? I’m also interested in the actual mechanics of executing this.
Of course. We decide the outline of the tasks. It’s just that most of the tasks I have seen were for 1 month.
Good point @hb1. This can drive more users for a longer duration. Even if someone misses first 3 rounds, they can submit in the fourth one!
The whole thing is about paying for tasks in Crypto to increase awareness, participation and finding new ways in the crypto ecosystem! We can also say “$2000 in NMR, top 20 participants get {(2000/NMR_price)/20}” but that can have price variations too.
First, we propose a task, if the funding is secured and task is accepted, we have to transfer the prize tokens to their address. They’ll distribute to the participants after the results.
@surajp Outreach is good, although I’m not familiar with 1729. Is there any public metric of their reach / audience composition to justify the specific choice of this platform over alternatives? And what would those alternatives be?
I participated in few tasks and learned a lot from them. They have 21.6K Followers on Twitter. This is an initiative by Balaji Srinivasan with 421.4K Followers on Twitter. A lot of whome are Crypto investors, AI engineers, traders, influencers. Who I think would understand the idea of project and probably would like to participate(We have more ways to participate now). I have observed their audience to be fairly technical people in crypto who can invest and probably stake.
One easy alternative I can think of is Kaggle(It was used for initial user acquisition I guess) but it’s limited to ML users only, not many of them are in crypto who’d buy NMR to stake.
Genius! I think 1729 is a great platform to bring Numerai under the attention of a new group of crypto and ML enthusiasts. Opportunity costs are very low here.
I do think keeping track of who these new sign ups are is important. Can we give them a referral code?
We’d also benefit from a clear definition of the task. The tournament is quite complex, so the simpler the better. I’m thinking something along the lines of “submit a new model and acquire a positive correlation in at least 1 round in the next month”.
Are all 1729 participants paid out, or just the first x that sign up or provide proof of completion?
Good points @jorijnsmit . Referral codes would complicate the process, I want it to be as similar as a new generic participant. We can use numerai_reports to see if the submission is from a new account or an older one and make sure that only one model is submitted from an account.
Based on the budget, we can pay more users. But I do like the sense of competition for a month.
as @hb1 pointed out, the top 20 of the best submitors in any 4 rounds will be paid $100 each. (This looks like a decent amount. Even if the winners want to withdraw the prize, they’ll have something after paying gas fees).
A similar task for signals is already in drafts. So if this works well, we can scale it next time.
This speaks to another issue. While greater reach is supportive for token demand, we (both Numerai and existing participants) truly don’t need any more defi degens. If / when the scoring moves to MMC or TC exclusively, then these initiatives should be pursued.
I like the idea. Also 2k seems like a nice starting point. Seeing as you also are rewarding for any model submission and not staked models, this initially wont effect the factor too much unless the users keep learning and participating down the road. Im not super worried about payout factor as much as others though. Signals is just sitting there with a nice 1 factor and a big pool still so I have a feeling the payout will level off at some point and more people will jump to signals. This could lead to another 1729 proposal down the road for a signals contest
Even if we keep the current pool of users and no new sign-ups, the payout factor will keep going down faster until it converges to 0.2 then it’ll decrease slowly.
Signals task is already in my drafts, we just need to see how much impact does this small experiment make. Then funding for the Signals task can be increased.
Yep, Ive seen a lot of users very worried about the payout factor in some of the projects proposed but fact is, even if nothing was done its going to keep dropping pretty rapidly
