The market is neither good nor bad, people’s hearts are neither ancient nor modern, Memes are born and die, and FHE comes to Privasea.
With the launch of the KOL matrix, I suddenly completed the re-education of the FHE track, and remembered the difference between FHE and ZK, and the combination of the DID track and NFT. However, none of this is important, because what retail investors care about is Binance's brand effect, the regulation of market makers, and the optimization of the Listing process.
Not to mention, Privasea is a representative project of the FHE+DID+AI track invested by Binance. Today I don’t intend to elaborate on how it is combined, it’s not important.
Image caption: YZi invests in Privasea, Image source: CryptoRank
Crypto + AI are just wishful thinking, and FHE and DID are not hot topics today. Putting them together can summon MKT magic, but just like Nillion and Arcium, they have not found their own PMF, except for tokens.
FHE track enters shock phase
I am not saying that there is something wrong with FHE technology, but that the projects on the FHE track in this cycle, including but not limited to Zama and Inco, have not explored the application paradigm in the Crypto field. I must repeat my previous point:
The combination of ZK and L2/Rollup is not a need for privacy, but ZK is naturally suitable for the "simple verification" scenario. When L2/Rollup transmits information to L1, the mathematical proof encrypted by ZK can be confirmed without fully revealing the original content. This is also the so-called difficult to calculate and easy to verify technical feature, rather than L2 can only use ZK technology, because the optimistic verification system can be operated using economic design (challenge period).
This is the real PMF of ZK technology in the Crypto field, which is to help L2/Rollup run more efficiently and securely. When we re-examine DID/TEE/FHE and Crypto + AI, you will find that apart from tokens, the products themselves do not have many usage scenarios and value. Even for the ZK hardware track, it is difficult to say whether it can develop in the future.
Returning to Privasea's solution design, everyone certainly has the need to verify their identity while protecting privacy (DID), and everyone certainly has the need for privacy protection and data encryption when using AI, but this does not mean that everyone has the need to use FHE.
In fact, since the birth of Zama and the huge amount of financing it received, the adoption of FHE in the Crypto field has focused on replicating ZK's L2 applications. Subsequently, many similar products were born around Zama's TFHE algorithm library. Without exception, they are just extras to a certain type of service.
Privacy is a Feature/Service, But Not a Product.
Privacy products are on the way - Signal
There are both good and bad news. Although Privases is not a surprise, Signal is really popular. At least now the US Department of Defense and Vice President Vance are using it. Compared with Telegram and WhatsApp, which both have E2EE (end-to-end encryption), Signal's end-to-end function is enabled by default, while the latter two require manual settings to be turned on.
But now, at least privacy products, which should have been centralized entities, have been favored by centralized departments because of their absolute security, which just proves the feasibility of privacy products.
What’s more worth mentioning is that Signal currently relies entirely on donations to operate, and they do not hope to transform into a new Meta or Google, but instead focus on maintaining their own operations and pushing the power of technology to the extreme.
Currently, it is estimated that by 2025, Signal's annual operating expenses will be as high as 50 million US dollars (estimated for 2023). For other types of products, the increase in user data will be synonymous with advertising revenue, but for Signal, it means a surge in server costs.
Image caption: Signal's spending structure in 2013, Image source: @signalapp
In 2013, the amounts for storage, servers, registration fees and bandwidth, as well as additional expenses, were specifically US$1.3 million, US$2.9 million, US$6 million, US$2.8 million and US$700,000. Among them, with the growth of users and the surge in data volume, even if Signal does not store user data permanently, it will require huge server and data storage costs. It is very difficult to predict whether these costs can be covered by donations.
If Signal can really run on donations, its significance will be no less than the birth of Bitcoin. Before this, technology donations were made by large companies to the underlying Internet protocols or underlying facilities such as Linux and Rust. There has never been a global C-end social product that can be maintained by small donations from ordinary users.
Looking forward to new history and optimistic about the implementation of privacy economics in Web2.
History will never end
Privases is just an introduction. It does not provide an in-depth analysis of the money-making effect, nor does it analyze its internal mechanism. Again, it is really not important.
What is really worth thinking about is that ZK has been successfully implemented on the B-side of Crypto, and end-to-end encrypted social products can maintain operation on the C-side. They have all passed their embryonic stage and entered the stage of large-scale practical use. What should FHE do?
FHE may not be the next step for ZK, but it is unlikely that FHE is not the next step for ZK.
Ultimately, the combination of FHE and Crypto is most likely to be in the field of private transactions. Only when the benefits of using privacy services are higher than the losses caused by complexity can FHE find its place.
For example, the Dark Pool that Arcium was considering yesterday is a subject of large-value transfers and transactions. Whether it is hackers who do not want to cause fluctuations in the secondary market or hackers who want to use special channels, there is a real demand for it.
In the transaction path supported by FHE, whether it is the transfer of 1 ETH or the transaction of 1000 WBTC/USDT, it cannot be "seen", which can fundamentally avoid the MEV problem. This is more scientific than Binance directly requiring BNB Chain nodes not to allow MEV.
Of course, if it were that simple, then Zama would have started a business in this direction a long time ago. Who would bother working on L2/Rollup? The biggest problem with FHE privacy transactions is that they cannot guarantee "identity". For example, if you trade on the FHE L2 chain, then you can indeed maintain privacy end-to-end.
However, once the FHE L2 transaction is sent to the Ethereum mainnet, it will face a dilemma. If you retain the encryption feature, the other party will see it as an indecipherable garbled code and will not be able to judge the transaction. If you remove the encryption feature, congratulations, there is no need to use FHE encryption from the beginning.
How to solve it? The only way is for Ethereum to natively support FHE opcodes directly from the consensus layer, which means that Ethereum will have to be modified to a certain extent, which is equivalent to the difficulty of converting PoW to PoS, and is basically not feasible.
This may not be the ideal transformation path, and FHE still needs to conduct its own exploration.
The only exception
If the Ethereum mainnet cannot be transformed, then guiding users to conduct dark pool-type and compliant privacy transactions (RailGun) is the only feasible implementation scenario. After the US government lifted sanctions on Tornado Cash, institutional compliance-level dark pools seem to have a glimmer of hope again.
After Privasea, FHE has rarely been used as a highlight for project financing, and FHE-AI/ML/LLM have all remained in the academic field. I firmly believe that blockchain, which originated from cryptography, is still the most suitable hotbed for FHE entrepreneurship.
