Waterfall х Crypto Arena AMA session May 21

Igor BerlenkoIgor Berlenko
30 min read

Sometimes, keeping quiet is the hardest part of holding on.

So, I keep my mouth shut and I listen, afraid I lose you if I didn't get we f for we're missing and compromise. Thus, I keep my silence, not leaving. We're in the nice winter season, and I can't wait to find a single reason not to say goodbye. Oh, if I said what's on my mind, would you stay with me tonight? If I said what was on my mind, would that make you run and hide?

Rolling high, if what was all on my mind would make you run and hide, oh, would that make you run? I don't know what to do anymore; I'm out of options. Nothing's working; I'm out of service. Please come help me now; I'm nervous. I'm trying but can't scratch the surface. So, I keep my mouth shut and I listen.

We realize that this approach is kind of... it's not a transition from zero to one. We decided to continue the research. I started my PhD; the topic of my PhD was scalability solutions with a focus on directed acyclic graphs. Essentially, Waterfall was the result of my PhD research. Currently, it is a result of the work of our entire research and development team, but the initial idea came from me. It has been improved multiple times on multiple levels by the brilliant colleagues I am proud to work with. Apart from that, I also do some work for IEEE and I'm advising the government of one of the countries on the implementation of their Web 3 strategy. This is it on my end; I'm passing it over to Y.

Thank you. Hello everyone, my name is Yan Leon, and I am an Associate Professor with a PhD in mathematics at the National University in Ukraine. I'm also working on IEEE projects. My main research purpose lies in mathematically modeling complex computer and economic systems. I've contributed to numerous international scientific studies in this field. For the last seven years, I have focused on the field of distributed ledger technology and have been involved in various projects concerning both decentralized public and private networks. I have worked at Waterfall for over two years in the research team, and I am pleased to be here to respond to your questions about our projects. Thank you for having me.

Thanks, please. Alex?

Hello, my name is Alexander Nan. I'm the development team lead on the project. I have a master's degree in computer science, and I have been working in development for 15 years. I have been involved with blockchain and other distributed ledger technologies for about seven years. I started working with and implementing his idea about Waterfall three years ago, and for that time, I have been working on it. I am also a member of the LTP group for current payment transactions. I have an open-source membership in this group. That’s a short introduction about me.

Got it, got it. That's a great introduction, and I'm happy to know about you all. Now, let's jump into the main topic: what is Waterfall? Please highlight some of its key features for our audience.

Sorry, you muted, sir.

Yes, thank you for pointing that out and for this question. Let me start sharing the screen. We will be very brief because there are multiple ways to describe Waterfall—the short one and the long one. We will try to stick to the short one.

The easiest way to understand what Waterfall is, is to review how our test net explorer works. This is a visualization of our test net; each of these drops is a block. You can click here, for example, and see the transactions and references to parent blocks, etc. Basically, what it allows us to do is have concurrent block production and, at a later stage, concurrent processing.

=> 00:10:41

Waterfall's parallel processing means no more waiting for transactions—experience true scalability without the delays.

Thank you for pointing this out, and thank you for this question. Let me start by sharing the screen. We will be very brief because there are multiple ways to describe waterfall; however, we will try to stick to the short explanation.

The easiest way to understand what waterfall is, is to review how our test net Explorer works. This is a visualization of our test net, where each of these drops represents a block. You can click on a drop, for example, and see the transactions and references to parent blocks, etc. Essentially, what this allows us to do is achieve concurrent block production and, at a later stage, concurrent processing, which enables us to achieve very high scalability while preserving all the standards of the Ethereum Virtual Machine. Additionally, it maintains very low node requirements and high decentralization.

To explore this further, let’s go to the dogs section of the waterfall network. Here, you can connect to the network using the RPC nodes. You can use MetaMask; essentially, you are able to receive your test net tokens. There are also tutorials available on how you can set up the apps. This is a one-click setup, although you will need to make more than one click in total. In general, it is a straightforward process, and you can also use Docker, which is more convenient for server versions.

You can proceed to use smart contracts with tools like Hardhat, Remix, and Truffle. There are examples for WC20, which is almost identical to ERC20, and WC721, which is identical to ERC721. If someone is willing to deploy a decentralized exchange (DEX), we currently do not have a DEX on waterfall. However, you can deploy your version of Uniswap V3 and improve it, potentially becoming the first DEX on our network if no one else has deployed theirs yet.

To clarify, when I say we do not have a DEX, I mean we do not have a community-owned DEX. To test the functionality of waterfall, we deployed Uniswap using its open-source code. We also did the same with Curve Finance. It is actually a pretty straightforward process because we are using an actual EVM. You can simply copy and paste smart contracts, although you will need to change some addresses and connect the front end. In general, this process is very different from developing a DApp from scratch.

Now, regarding our testing results, we started our first test net in October of 2021 and presented it in Dubai. At that time, we reached around 1,000 transactions per second. In the second test net, we achieved scalability of around 2,000 transactions per second. There were multiple test nets, including test net three and test net seven. Currently, we have a public test of test net seven, where someone, not us, sent 2 million transactions to the test net. The network handled it without delays, which is impressive.

What is particularly noteworthy is that these transactions were included in one block in the slot, while the rest of the transactions were processed in parallel. This means we did not experience delays for transactions happening concurrently with the high volume. For example, if you remember the time with Ethereum when they released CryptoKitties, you could submit a transaction in the morning, but it might be processed in the evening due to the linear structure of the blockchain. In our case, because we have parallelism, even if there is a long queue of transactions, we can still process them concurrently by adding them to other blocks.

By the way, during that time, some other protocols were also being tested with a similar number of transactions, and they experienced delays. This highlights the advantage of utilizing a Directed Acyclic Graph (DAG).

The last test is actually test net eight, where we were able to...

=> 00:15:40

Scalability meets decentralization with the next-gen directed acyclic graph technology, paving the way for faster transactions and community-driven networks.

In the morning, you could have S transactions processed if they are good, and this would be completed by the evening. This efficiency is attributed to the linear structure of the blockchain. In our case, however, because we have parallelism, even if there is a long queue of transactions, we are still able to process transactions concurrently by adding them to other blocks. At that time, some other protocols were experiencing delays while testing, which highlights the advantage of utilizing a direct acyclic graph.

The last test conducted was on test net 8, where we achieved more than 11,000 transactions per second. I would also like to point out the node requirements: we are using two virtual core CPUs and 4 GB of RAM. Essentially, one virtual core CPU equates to one physical core CPU. For example, your smartphone typically has between 4 to 8 core CPUs and 8 to 32 GB of RAM. Therefore, our node requirements are significantly less demanding than those of a smartphone. It is important to note that while the cores of a smartphone are not the same as those of a server, there is a basis for comparison when considering one physical core here and four physical cores there. This is why we are optimistic in stating that at some point, waterfall is planned to work with mobile devices, including smartphones.

Speaking a little about traction, we have dogs and waterfall that Network. This is the page of statistics where you can find the total number of workers. A worker is a block-producing unit, and currently, we have 2,658 workers. In parentheses, this indicates the number of workers added in one day, which is six. Out of these, 248 workers are controlled by the development team, and we are disclosing this information. The rest, which is already more than 600, is controlled by the community, including early adopters, testers, and developers. This means we have 600 block producers controlled by the community.

Additionally, you can find links to the one-click setup node and the reward ratio on the page. However, it is important to clarify that this is a non-incentivized test net, meaning the rewards are essentially worthless and will not be traded on exchanges or utilized on the main net. Despite this, we appreciate the community's attention and are striving to deliver a high-quality project for the community to start using soon.

I believe this covers the main points, and we are happy to address any further questions you may have.

Thank you so much for such a wonderful presentation. I think all the listeners have enjoyed hearing about waterfall so far. I have a list of interesting questions that I would love to know the answers to. One of the questions is about the waterfall and its directed acyclic graph framework. Can you explain how the DAG is integrated with the blockchain and what specific security benefits it offers compared to traditional blockchain approaches?

Thank you for this question. Let me pass it over to Yen, who can provide insight into how we are using DAG on the Shar Network and how we plan to utilize it on the coordinating network. After that, I will add something to this question.

Can you see my screen with the dark figure? Yes, we see it. Great! The DAG Ledger design can be considered the next generation of blockchain technology, capable of solving the scalability problem while preserving decentralization, unlike traditional blockchain systems.

=> 00:21:13

The future of blockchain is here: a scalable, decentralized solution that harnesses the power of parallel processing to redefine efficiency.

The discussion begins with a focus on how the G is integrated with their blockchain and the specific security benefits it offers compared to traditional blockchain approaches. The speaker expresses gratitude for the question and then hands over the conversation to Yen, who will elaborate on the use of D in the Shar Network and the plans for its application in the coordinating network. After Yen's explanation, the speaker will add further insights.

Yen then shares his screen, displaying a dark figure, and confirms that the audience can see it. He explains that the duck Ledger design can be considered the next generation of blockchain technology, capable of solving the scalability problem while preserving decentralization. Unlike traditional blockchain technology, which relies on linear sequenced blocks in a block database system, the timeline in their system is composed of distinct slots. Each slot can contain multiple blocks, as depicted in the figure. These blocks form referential structures since they must reference several previous blocks on the waterfall platform and other similar networks.

As a result of operations, the created blocks of transaction line up in a part of the deck that is topologically sorted and finalized. This finalized part of blocks is referred to as the "spray," since finalized blocks of the same slot are created in parallel. This approach significantly impacts the current level of scalability, helping to solve the scalability problem and allowing for a higher transaction throughput. Yen thanks the audience for the question and then passes the floor back to the speaker.

The speaker acknowledges Yen's explanation and adds that they are also working on implementing D in the coordinating network, which would allow for virtually unlimited sharding. Yen is invited to elaborate on this approach. He explains that their research team has recently completed a new method that allows for virtually unlimited sharding on the waterfall platform. This method aims to overcome scalability limitations while preserving full decentralization. The results have been submitted for publication, and the team continues to work on system design and new features.

The main idea is that the coordinating network operates as a deck, just as each shard network also has a deck structure. In simpler terms, they are utilizing DAGs on two levels, which brings the concept of parallelism into play. This parallelism is implemented at multiple levels: within the coordinating network and the shard network. This structure ultimately leads to parallel processing on the shard level and extends to parallel processing on the multi-shard level. The contribution of the DAG is crucial in pushing the boundaries of scalability forward.

The speaker encourages the audience to refer to the documentation, white paper, and website for deeper information about the waterfall. Moving on, another question is directed to Mr. Shi regarding the low minimum hardware requirements emphasized by waterfall. Some may view this as a step towards dedicated proof of stake, and the speaker asks Mr. Shi to address this concern while elaborating on how waterfall maintains a balance between decentralization and efficiency.

Mr. Shi responds by thanking the speaker for the question. He clarifies that waterfall operates as a pure proof of stake system, which may be a bit confusing. In this model, anyone can have a node, and each node has a fixed amount. This structure is similar to Ethereum, where a limited number of people do not control the nodes while others stake their coins on behalf of those nodes. He further explains that they do have a delegation mechanism, but this delegation...

=> 00:27:08

Waterfall's proof of stake empowers everyone to run a node, striking the perfect balance between decentralization and efficiency.

Waterfall is designed as a pure proof of stake system, which means that it does not rely on a limited number of nodes operated by a select group of individuals. Instead, everyone can have a node, and each node has a fixed amount. This model is quite similar to Ethereum.

In addition to this, Waterfall incorporates a delegation mechanism, primarily for security purposes. For instance, if an individual possesses a large number of Water coins for various reasons—such as liquidity or other use cases—they may want to stake these coins without transferring them to hot wallets. This is where the delegation mechanism comes into play. The individual can post a staking transaction that utilizes the coins stored in their cold wallet, allowing them to stake on multiple servers and create multiple workers on each server.

This system is also beneficial for infrastructure providers, who offer node infrastructure services to those who prefer not to host their nodes themselves. It essentially enables a non-custodial staking process, where the coins are delegated to a server to produce blocks, and the rewards are distributed accordingly, provided it complies with the laws of their jurisdiction. Thus, in Waterfall's case, delegation serves as a security measure.

Looking ahead, there will be numerous announcements regarding partnerships with various node infrastructure providers aimed at enhancing user experience and the quality of servers available to the community. This is particularly relevant for users who may not wish to store their nodes on their personal computers.

Moving on to another question, Mr. Alexander was asked to elaborate on the technical trade-offs involved in this approach and how it impacts network security and decentralization compared to more resource-intensive full nodes. However, there was some confusion regarding the phrasing of the question.

To clarify, it was suggested that the question be rephrased for better understanding. In the meantime, another inquiry was raised about whether there is a white list for buying tokens. The response indicated that the technical team primarily focuses on technology development, and for questions related to purchasing coins, individuals are encouraged to reach out to Richard Wang, a general partner at Dragon Fund, or Brett, the investment director for Waterfall. They can also contact the investment relations team at investrelations@waterfall.network for further assistance.

In conclusion, the previous question about decentralization and the implications of the project was acknowledged as interesting, and it was suggested to rephrase it for better clarity.

=> 00:32:38

Decentralization is key to a secure network; the more diverse the nodes, the harder it is to control.

In our ongoing efforts, we focus on technology development while encouraging people to join our communities. Interested individuals can reach out to Richard Wang, a general partner at Dragon Fund, or Brett, who serves as the investment director for Waterfall. They are available to answer questions, and inquiries can also be directed to invest relations at waterfall.network or ir.waterfall.network.

Regarding a previous question about centralization, I would like to clarify some points. The question I received in text form was somewhat unclear to me. It pertains to decentralization, and many individuals working on decentralization projects believe that with increased decentralization, we should have more servers or nodes with high requirements. We utilize parallel processing for many tasks, where some tasks operate independently while others do not. For example, within our committees, workers send each other attestations about information. This information is then aggregated and sent to our Global Network.

We break down algorithms into smaller parts, allowing different nodes to handle specific tasks. The results are then compiled and verified by global members within the network. Additionally, we have a mechanism in place to address situations where a worker may not perform their duties correctly. If a worker makes a mistake, other workers can report this, and the underperforming worker will face penalties. This approach enhances the security of our network.

Thank you for highlighting our solutions aimed at increasing security within Waterfall. Moving on, I have another question regarding the measures Waterfall has implemented to ensure maximum decentralization. There are several ways to assess this. One method is through the Nakamoto coefficient, which represents the number of entities that can collude to disrupt the network's operation.

This can be examined from various perspectives. For instance, if we have numerous nodes all hosted on Amazon, the Nakamoto coefficient would be one. If the nodes were distributed across Amazon, Google, and Microsoft, the coefficient would rise to three, as these three entities could potentially collude to affect the network's functionality. A similar situation occurred in Bitcoin around 2017-2018, where the top three mining pools controlled over 51% of the network and were primarily located in China. In that scenario, the coefficient could be one if the Chinese government decided to shut it down, or three if all three pools colluded.

Moreover, it is essential to consider ownership concepts. If nodes are diverse but owned by the same entity, it still reduces the Nakamoto coefficient. Our goal is to create a highly decentralized technology that combines scalability and decentralization, potentially involving millions of block producers. We are also collaborating with KYC providers to facilitate ownership verification, which will be integrated into our Decentralized Autonomous Organization (DAO).

On our website, we promote the principle of one person, one vote. In this context, the DAO is designed to function democratically. Typically, some DAOs operate on a model where individuals with more coins have greater voting power, which can lead to a governance structure resembling that of an enterprise. In an ideal world, however, we strive for a system where voting power is equitable, regardless of the number of coins owned.

=> 00:38:41

Decentralization and scalability can coexist, empowering everyone to have a voice in a democratic system, not just the wealthy.

To combine scalability and decentralization at the level of having millions, and potentially more, blog producers, we are also working with KYC providers. This collaboration will later enable the verification of ownership. Furthermore, it is going to be involved in the DAO (Decentralized Autonomous Organization). On our website, we have implemented the principle of one person, one vote. In that regard, the DAO is planned to operate in a manner that differs from traditional systems.

Typically, some DAOs operate on a model where a person who owns more coins has a higher voting power. This can lead to a scenario resembling the government of an enterprise, where the more voting shares one possesses, the stronger their vote becomes. However, when discussing democracy, in an ideal world, it should not matter whether one is rich or poor; everyone should have one vote and be able to use it equally.

In the context of democracy, we also have what is known as delegated democracy. This means that individuals do not vote on every decision made by the government but rather elect representatives who make those decisions. Depending on the specific case, we plan to implement various levels of understanding this from the perspective of game theory. For example, directed voting may relate to grants, allowing people to vote on which applications they will use.

In more complicated matters, such as protocol upgrades, there should be community voting alongside an expert committee that makes the final decisions. This is important because implementing direct democracy can make the system vulnerable to issues such as the tyranny of the majority. Additionally, if the majority lacks expertise, it can hinder the decision-making process. Our goal is to achieve a very high level of decentralization, ensuring that not only highly technical individuals own nodes but that regular people also contribute to the security of the network.

While this aspect is not yet strictly decided, we are committed to pushing it significantly further. With the implementation of mobile devices, we can potentially reach even higher numbers than those previously mentioned. The challenge we face is how to combine scalability, sufficient for mass adoption, with decentralization that can accommodate a large number of independent participants.

I would like to emphasize once more that preserving decentralization is crucial for us, not just achieving high transaction throughput. Everything our team does is based on the assumption of maintaining a very high level of decentralization in our system. To achieve this, we are developing a new fast finality proof of stake consensus protocol that involves the participation of millions of nodes. Additionally, we are focusing on scalability in terms of transaction processing capacity and working on parallel processing, routing execution, data traffic, and data storage.

Now, regarding whether beginners can set up a test node or if it is more suited for advanced users, I will share my screen to demonstrate.

As I share my screen, you can see our website. Yes, we are able to see that. You can download our application and add a node in a few ways. You can go to our website and click on "Join Testnet," where you will find the download link at the bottom. Additionally, you can refer to the documentation on our site, which includes tutorials for downloading the application.

We offer two versions for installation. The first option is to install Docker and launch our application or nodes within Docker. This option is more suited for professional users, but we provide comprehensive instructions that many beginner users have successfully followed to install and run our node.

The second option is to download our native application called One Click Setup. This application is available for your computer, whether it is Intel or Mac, or Windows. After downloading, you can launch the application, and it will set up everything for you.

=> 00:44:38

Setting up your application is easier than you think—just follow the steps, choose your setup method, and remember to keep your mnemonic phrase safe.

You can go to the documentation for our site, where we have tutorials available. Additionally, you can download the application; we have two versions to install. You can install Docker and launch our application using Docker. This option is more suitable for professional users, but we provide good instructions, so many beginners can follow these steps to install and run our application.

In the second way, you can download our native application called On Click Setup. You can download it for your computer, whether it’s Intel, M1, or Windows. After downloading, you can launch the application, which is our fixed-up application.

Now, I have one note, but the application hasn’t launched yet. First, you should note and select the type; now, the local node is located in the local database on your disk. You must use an SSD for this. Next, select the network; we have one T net eight, which is the location for the data folder where you want to save the database.

For example, you have two options next. You can check the box to download our latest snapshot, which needs about 70 GB of space. However, you will need more space because you have to download this snapshot and then extract it. In total, it will require about 80 GB. If you don’t check this box, you will synchronize from Genesis, which can take a lot of time—maybe three days or more. This week, we will check how much time is needed for synchronization from Genesis. If you don’t have time for that, you can download the snapshot from us. We deploy this snapshot about once a week and upload it to Google Drive, so you can download it from there.

After that, you should change the port because I already have one node running, and this port is in use. You can see that you should change it to avoid conflicts if you use the same port for two nodes in the application. After that, you should set the name for the node and provide the necessary information. Then you can create the node, and now we have the status showing that it is downloading the snapshot. After the information is empty, it means the node doesn’t run yet.

The second step you should take is to add a worker. We also have next steps illustrated in the instructions, which include screenshots. I know we have a video about it, but I’m not sure if it has been uploaded to YouTube yet.

Returning to our application, you can go to the workers for this node. You can import from another computer or add a new worker. I recommend doing this in the workers section of the notes to avoid mistakes regarding which notes you are using. You should select a worker and regenerate the mnemonic. It is crucial to save this mnemonic in a safe storage location.

Music plays

You should never show your mnemonic phrase on the screen to anyone—never ever show your mnemonic phrase to anyone. Sasha Alexander is demonstrating how you shouldn’t use your mnemonic phrase. Now, he is switching to show how you should do it correctly.

After verifying your mnemonic, you can choose how many workers you want to add—maybe two or three. You will also need to provide your withdrawal address. I will check an example for that. This withdrawal address is linked to your mnemonic phrase, which needs to generate keys for your workers. These workers should be set up accordingly.

=> 00:51:49

Don't let your mnemonic phrase slip; keep it private and secure to protect your assets.

Sasha Alexander is currently demonstrating how you shouldn't use your mnemonic phrase. He emphasizes that you should never show your mnemonic phrase to anyone. He quickly illustrates the incorrect way to handle it, and then transitions to the correct approach. He reiterates, "don't show your mnemonic phrase to anyone."

After verifying the mnemonic, you can choose how many workers you want to add—perhaps two or three. You will also need to provide your withdrawal address. For example, he mentions, "I will use this withdrawal address." The mnemonic phrase is essential as it generates keys for your workers. These workers are responsible for creating blocks, signing them, and establishing the necessary connections. If you encounter problems with any workers, you can deactivate them and create new ones.

In the event that you lose your mnemonic phrase, you can still add new workers, but you will need to deactivate the existing ones. You will use the withdrawal address for this process, allowing you to remove workers from the network and create new ones. While losing your mnemonic phrase is indeed a problem, it is not a security problem.

Currently, two workers have been added, and the node is still synchronizing. Once synchronization is complete, you will be able to enable the workers and send transactions to activate them. Sasha concludes by stating, "that's all I want to show about for beginners." He believes these steps are not difficult and encourages the creation of tutorials to help people interact more effectively, particularly regarding how to create nodes.

Moving on to another topic, a question arises about how waterfall decentralization compares to additional blockchain models. Sasha returns to the screen to provide a visual explanation. He describes a traditional blockchain as a straight line where one block follows another. In contrast, the waterfall model allows for parallelism, enabling the creation of a wider sub-network concept. This concept allows nodes to be split into strongly connected subgroups that can create one or multiple blocks concurrently with other subgroups.

For example, nodes can be located in the United States, India, China, and Europe, with each group creating blocks that merge into the same chain. This parallel processing on a single shard level accommodates a large number of nodes while preserving scalability and maintaining low node requirements. Sasha believes that DAG (Directed Acyclic Graph) technology is the future of distributed ledger technologies.

He explains that, like blockchain, DAGs possess the essential qualities of scalability, security, and decentralization, which often present a dilemma between scalability and decentralization. He notes that if security is lacking, the technology cannot be trusted. There are different types of security, including probabilistic and deterministic, and he mentions that the waterfall protocol combines both.

Sasha expresses high confidence in their ability to push scalability and decentralization to levels that are significantly different from what is commonly found in the market. He concludes by stating that they are committed to advancing these capabilities in the future.

Finally, he addresses a question regarding how energy-efficient the waterfall protocol is compared to other protocols.

=> 00:57:30

Balancing scalability and decentralization is key to achieving energy efficiency in blockchain technology.

There seems to be a dilemma between scalability and decentralization. If you don't have security, you don't have Ftech. Obviously, there could be different types of security, such as probabilistic or deterministic. By the way, waterfall combines both of them. We have a high level of confidence that we have already been able to push scalability and decentralization to a very different degree than what is commonly seen on the market. We are committed to increasing this and pushing it forward.

Thank you for this question. Now, I have a last question for you: How energy efficient is the waterfall protocol compared to the other block DAG protocols? Let me put it this way: when speaking of energy efficiency and node requirements, they are somewhat similar. Generally speaking, there are good block DAGs, but those working on proof of work consume more energy. For example, Bitcoin consumes more energy than protocols using proof of stake.

We are using proof of stake with low node requirements, designed to accommodate a large number of nodes. Proof of stake is generally considered energy efficient, and we are pushing the node requirements lower to decrease the energy requirement for the protocol to function. All of these measures may sound a bit complicated, but they are moving in the direction of achieving low node requirements, high decentralization, and high scalability simultaneously.

If you review the average number of nodes on the protocols, you will see some interesting statistics. Let me show you on the screen.

One moment, please.

Here we go. This is the number of block producers on popular networks. The first two have more than 100,000 nodes; Ethereum is almost at a million, while Bitcoin has less than 50,000.

I don’t want to say anything because I might say the wrong words. So, I’ll keep my mouth shut and listen. I’m afraid I’ll lose you if I don’t get it right. We’re missing a compromise, so I’ll keep my show and not leave. We’re in a nice winter season, and I can’t wait to find a single reason not to say goodbye.

If I said what’s on my mind, would you stay with me tonight? If I said what’s on my mind, would it make you run and hide?

I don’t know what to do anymore; I’m out of options. Nothing’s working. I’m out of service. I’m nervous, trying but can’t scratch the surface. So, I’ll keep my mouth shut and listen. I’m afraid I’ll lose you if I don’t feel we’re missing a compromise. We’re in a nice winter season, and I can’t wait to find a single reason not to say goodbye.

Okay, sir, could you please start again with the same presentation?

Sounds good. Basically, we were addressing the number of block producers for popular protocols. For example, Ethereum is close to 1 million, OSIS has more than 100,000, and Bitcoin has around 50,000. All the rest are measured in thousands, and the numbers continue to decrease as we go down the list.

=> 01:04:09

Embrace the challenge of balancing scalability and decentralization in tech—it's the key to building a thriving ecosystem.

Hi, roll it.

Let's begin.

Could you please start again with the same presentation?

Sounds good.

Basically, we were addressing the number of block producers for popular protocols. For example, we have Ethereum, which is close to 1 million, EOS is more than 100,000, and Bitcoin has around 50,000 nodes. The rest are measured in thousands, with the eighth protocol already in the hundreds. Going forward, the numbers drop to tens, such as 1,763, 58, etc.

What this means is that in most cases, protocols are making choices in favor of scalability with some sacrifice to decentralization. By the way, I'm not saying these are bad technologies; they are all good technologies. However, what we're trying to achieve is both high scalability and high decentralization, measured in millions and possibly more.

As long as we are already looking at my LinkedIn, I would like to share one important announcement. We are preparing for the launch, possibly even next month, and we are forming the ecosystem. We already have close to 10 partners who have signed up, and we are seeing good traction on this topic.

We are inviting bridges, oracles, digital identity providers, liquid staking, staking infrastructure providers, access lending protocols, gaming apps, stablecoins, and wallets to partner with us. Other applications are also welcome. Please join our community channels and follow the announcements. We look forward to having you all at our launch, and we hope you will join the exciting journey of building the waterfall and its ecosystem.

Alright, I think we have covered most of the project, and it’s time to wrap up. Before we conclude, do you want to say something to our community or offer a closing statement?

Let me go ahead then.

We are inviting everyone to join the test net and prepare for the launch. If you experience any bugs or encounter any inconveniences in the process, please report back to us because this way, you are contributing to the development of the project.

If you are a developer looking to deploy your dApp on new technology or a new protocol, please go ahead and do that. We currently have around 45,000 people across our community channels, and this is not a community of fake users. These are users who are highly engaged and enthusiastic about our technology.

We welcome collaboration opportunities, and everyone is welcome to join. Thank you so much.

Please support the waterfall project and be part of this community. If you have more questions or need assistance with anything related to the waterfall, drop your questions on their social links. Their community manager or lead member will definitely reply to you.

That was it for today. I’ll see you next time. Have a great day!

Bye-bye, and thank you for joining us.

I don't want to say anything because if I do, I might say the wrong words. So, I keep my mouth shut and listen, afraid I might lose you if I didn’t.

We are missing something and compromising, so I keep my show on. Not leaving, we are in a nice winter season. I can't wait for a single reason not to say goodbye.

If I said what’s on my mind, would you stay with me tonight?

If I said what’s on my mind, would that make you run?

I don’t know what to do anymore. I’m out of options; nothing’s working. I’m out of service. Come help me now; I’m nervous. I’m trying but can’t scratch the surface.

So, I keep my mouth shut and listen, afraid I might lose you if I didn’t feel a way. But we are missing something and compromising, so I keep my show on. Not leaving, we are in a nice winter season. I can’t wait for a single reason not to say goodbye.

If I said what’s on my mind, would you stay with me tonight?

If I said what’s on my mind, would that make you run?

I don’t know what to do anymore. I’m out of options; nothing’s working. I’m out of service. Come help me now; I’m nervous.

I’m trying but can’t scratch the surface. So, I keep my mouth shut and listen, afraid I might lose you if I didn’t feel a way. We find we are missing something and compromising, so I keep my show on. Not leaving, we are in a nice winter season. I can’t wait to find a single reason not to say goodbye.

0
Subscribe to my newsletter

Read articles from Igor Berlenko directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Igor Berlenko
Igor Berlenko