Now that the claiming period for the first phase of Starknet Provisions (‘airdrop’) has ended, I want to share my personal impressions. None of this is investment advice, nor does it necessarily reflect the position of StarkWare or of the Starknet Foundation; do your own research.
What is Starknet and Why STRK?
Recall that Starknet is a validity rollup (‘zk rollup‘) launched in Alpha mode in November 2021. It scales Ethereum via cryptographic protocols called STARKs, and does so without compromising Ethereum’s core principles of broadness (‘decentralization’), transparency, inclusivity and security.
The Starknet token empowers people who contributed to the success of the ecosystem to play a role in the governance and ongoing operation and safe-guarding of Starknet. The token is to be used for (i) governance, (ii) payment of transaction fees on Starknet, and (iii) participation in Starknet’s consensus mechanism.
The Starknet Foundation allocated STRKs via a number of programs to valued community members who have proven they want to advance, maintain and secure Starknet. Some of the programs used include Devonomics, Catalyst, DeFi Spring and Provisions – the focus of this post.
On February 14, 2024, the Starknet Foundation announced the first round of this program, allocating a total of up to 700+ million STRK tokens out of 900M STRK earmarked for ‘Provisions’. The claiming period lasted four months, between February 20 and June 20, 2024. By the end of it, roughly 500M tokens were claimed, leaving roughly 400M STRK tokens for future rounds.
What was the goal of Provisions?
The main goal of Provisions was to allocate STRK tokens to a broad set of individuals, real people, who will operate the system and contribute to its security and governance, contributing to its decentralized nature. Starknet is a social tool – a technology for empowering individuals and societies to implement on top of it any social function they need, like money, property management and voting systems. Therefore, the security of Starknet is directly linked to the broadness and resilience of the set of people who care about it.
The big challenge faced by Starknet Provisions is that blockchains don’t understand people. By this I mean, the atomic units on blockchains are account addresses, not people or users. There is no clear correspondence between the two and one person may control multiple addresses. It is hard to gauge, based on the information present onchain, which accounts represent people who are more likely to contribute to the future operation, security and governance of Starknet. In other words, our problem is: How to allocate tokens to people in correlation with their long-term alignment with Starknet’s mission, given the data present onchain and elsewhere about accounts and their activity?
The first thing that’s clear to me is that the data at hand is simply insufficient and inadequate to solve the problem accurately. It was evident to all involved in Provisions from the get go that all calculation methods considered are mere approximations, and that the Provisions program would potentially miss the mark in certain cases. There will be users who fully align with Starknet’s long term mission and will receive a small allocation of tokens, or even no tokens at all, and there will be users that are not aligned with the mission and nevertheless end up with a lot of tokens.
It is important for me to state this in light of the public criticism that Starknet Provisions has faced, and which subsequent airdrops like Eigenlayer, ZKsync and LayerZero, have also faced. I am not aware of any available tools that would achieve the goals described above in a way that is considerably more accurate or fair than the methods we used. The allocation method wasn’t perfect, but other allocation metrics would have led to a different set of inaccuracies.
How did the Starknet Foundation approach the allocation problem?
As publicly announced, the approach taken by the Starknet Foundation was to allocate separately to six different classes of people, and the allocation within each class was based on metrics/data relevant to that class. Briefly, the classes and metrics used were:
- Starknet users. A metric of activity and usage was employed, and eligible addresses were passed through a sybil filtering system run by an external party.
- STARK early adopters. Users of the STARK-proof technology that predated Starknet (StarkEx users) were allocated, based on their usage.
- Ethereum contributors. Individuals who contributed to Ethereum in a variety of ways – by staking, developing code, participating in the builders guild and/or submitting Ethereum Improvement Proposals (EIPs) were allocated, with specific metrics employed for each sub-category.
- Github devs. Developers of select open source code projects on Github were allocated based on a metric that measures Github activity.
At the same time that people were allocated tokens , two additional classes of people received tokens, outside provisions:
- Early community members program (ECMP). Individuals who contributed to the Starknet ecosystem by running events, advancing Starknet’s community were allocated tokens by applying in advance. A committee of ecosystem members decided on the allocation based on reviewing the applications.
- Developer Partnerships (DPs). Developers of the core infrastructure of Starknet with pre-existing agreements with the Starknet Foundation were allocated tokens for claiming on the same day that Provisions claiming started. These allocations were agreed upon between the Starknet Foundation and the developer teams in advance.
The underlying theme here was to try and reach a diverse set of individuals who, judged by their past actions and contributions, are deemed a natural fit for the task of operating, caring about, and securing Starknet.
Did Provisions achieve its goal?
Given the inadequacy of metrics, as explained above, it was clear to us from the onset that some approximations would lead to inaccurate allocations. This raises the following question: Given the data sources at our disposal, did we do the best possible job? How do we assess this question? A related preliminary question is to what extent do recipient addresses match real people, i.e., to what extent do addresses correspond to individuals?
Of the six classes of people mentioned above, the last three do line up with real people, and we can even go further and speculate that those people are likely to care about Starknet. Of the third class (Ethereum contributors), all but stakers likely fall under “one address, one person”, and their past actions have shown that they care about decentralized systems, so one may hope they’ll care about and help secure Starknet as such. The class of StarkEx users, which are early adopters of the STARK technology, is the smallest class in terms of claiming rate and size (only 2.4M STRK were claimed, less than 1% of the total sum distributed). The class for which all the questions above is hardest to answer is that of Starknet users. This is the largest recipient class, receiving over 87% of the total Provisions claimed (over 430M STRK). And most of the public outcry that followed Provisions was related to the allocation to people in this group.
Social media, and in particular Twitter, had a lot to say about this. Most of it was scathingly negative and focused on the criteria for receiving an allocation as a Starknet user – requiring that at least 0.005 ETH be held in an account on a certain date. But other grievances surfaced as well. Two notable themes were the reposting of a heated comment made several months earlier by one of StarkWare’s senior managers, for which he publicly apologized shortly after, and an unlocking schedule for StarkWare shareholders (which includes investors, founders and employees) that was criticized as too quick. The unlocking schedule for shareholders was changed, and the intensity of criticism related to it subsided. But the other two themes – the 0.005 Eth cutoff criteria and the “e-beggar” criticism remained for many weeks and months. Both have subsided considerably, especially now that more recent airdrops are getting criticized, but they linger to this day.
What should we make of this outrage? To what extent does it come from farming teams who are rationally trying to influence the criteria for this round and for other rounds (of Starknet and other projects)? How much of it represents a broad set of people (“farmers” or “non-farmers”) who, under a different allocation method, would have contributed to the long-term success of Starknet? These are research questions that I’d love to see answers for. If you have a way to work on this, please post your suggestion on the Starknet Community Forum and tag me there.
So far I’ve discussed social media, related to the allocation to Starknet users. But let’s look at the bigger question. Did Starknet Provisions do a good job? The short answer is that I don’t know. The reason is the lack of metrics needed to answer the problem. This is the same issue underlying the inability to accurately allocate tokens to people in the first place. The metrics available onchain, like daily TPS, TVL, STRK price and #addresses do not directly answer the question: “are STRK holders today a broad and diverse group of people that will stick around to improve, operate and secure Starknet?”
I would love to get an answer to this question as well – if you have an idea on how to go about it, please post it on the Starknet Community Forum and tag me.
How did it feel?
A strange question to ask, but I believe many are curious to hear the answer. The emotional stress associated with the Provisions effort was extreme on the whole team but especially on those, like Abdel and myself, who were the focus and target of personal attacks.
To cope with a twitter feed filled with “scammers” (and worse), we collectively embraced each other, not just within the Starknet Foundation, or the StarkWare team but relied on the unrelenting support of the amazing ecosystem that is Starknet. While challenging to navigate, this period of scrutiny ultimately proved valuable. It highlighted areas for improvement and tested our resilience as a team. We learned the importance of standing firm in our decisions while remaining open to constructive feedback, even when it’s delivered harshly. This experience reinforced our conviction that in the crypto space, one’s approach to handling public pressure can be as important as technical decisions. It was amazingly empowering to see folks from other ecosystems (sometimes even our competitors) reach out and offer support, and I will never forget this. Most of all, we derived our strength from the amazing Starknet ecosystem.
My personal lessons for the future
Roughly 400M STRK remains earmarked for future rounds. What can be done better?
- It’s abundantly clear that “proof of personhood” is a truly hard problem to solve on blockchains. We can’t be sure that we will ever solve it but it is the kind of science and technology research that we passionately want to pursue.
- The incentive of farming entities to influence such rounds (or future allocations of other projects who are watching what we’re doing) remains as strong as ever, which means that whatever is done will lead to a public outcry on social media. To me, this is an unavoidable and unpleasant part of the business we’re in.
- I hope the Starknet Foundation and the Provisions team under it will find new ways to allocate tokens to a diverse set of people who care about the long-term vision and mission of Starknet and will stick around to secure, govern and operate it. I know this is their intention and they are conducting research and discussions on ways to achieve this.
Summing up, Provisions is about placing Starknet tokens in ‘the right hands’, of people who have demonstrated they care about, and align with, the long-term mission and vision of Starknet. I honestly don’t know whether the allocation methods used by the Starknet Foundation, especially regarding Starknet users, were adequate and sufficiently accurate. I hope that future research by the community will answer this question. I’ll certainly keep thinking about this and plan to share my suggestions at a later point in time.
I’d love to hear more thoughts on token distribution mechanics from people across the ecosystem and beyond, please post them on the Starknet Community Forum.
————-
THIS IS NOT AN OFFER TO SELL OR THE SOLICITATION OF AN OFFER TO PURCHASE ANY TOKENS, AND IS NOT AN OFFERING, ADVERTISEMENT, SOLICITATION, CONFIRMATION, STATEMENT OR ANY FINANCIAL PROMOTION THAT CAN BE CONSTRUED AS AN INVITATION OR INDUCEMENT TO ENGAGE IN ANY INVESTMENT ACTIVITY OR SIMILAR. YOU SHOULD NOT RELY ON THE CONTENT HEREIN FOR ADVICE for advice of any kind, including legal, investment, financial, tax or other professional advice, and such content is not a substitute for advice from a qualified professional.
