Back to Badlands: Rebuilding the Internet Computer
This is one of a series of posts that have been sitting in a perpetually open Chrome window on my desktop for the last 4 months. It was intended to be published in the context of some other ideas. However, when I have raised the thoughts contained here to people in the DFINITY community the reception has been so positive I thought I may as well just get this finished and out on Nuance (syndicated through other channels). Given it’s a sprawling and ever changing subject matter, there are inevitably some factual inaccuracies here.
This is an attempt to expand on Dom’s Badlands post from last year and takes inspiration from Albert O. Hirschman’s book Exit, Voice, and Loyalty. The book describes two options a person has when an organisation or institution doesn’t meet their requirements: Voice their disapproval in an attempt to influence positive change, or Exit by ceasing to rely on it, which might involve adopting an alternative. Loyalty represented a threshold for action.
This framework can be applied to blockchains and the Internet Computer: If you are unhappy with a traditional blockchain, recourse to Voice is difficult, but Exit is possible by hardforking to create a “new” blockchain and steering it in a preferable direction. We saw this with the Ethereum Classic fork and the range of SegWit forks of Bitcoin. In the case of the IC, the NNS makes it much easier to drive change through Voice, but Exit through state or client hard forking is apparently impossible. This issue was actually a subject of discussion back when the NNS (then BNS) was first publicised. The contention of this post is that Exit is indeed possible in the Internet Computer ecosystem. Dom’s Badlands post was a first step in sketching what this could look like.
This post is also inspired by the successes of the Kusama network. A network based on Polkadot:
“Kusama is wild and fast; great for bold experimentation and early-stage deployment. Polkadot is more conservative, prioritizing stability and dependability.”
While Dom’s conception of Badlands focused on democratising the running of nodes through reducing barriers to entry, specifically reducing node hardware costs, there are more opportunities a Badlands-type network could address to benefit the Internet Computer (henceforth IC) ecosystem. The first section of this post will define some constraints to consider in executing a new-network strategy, the second will look at opportunities a reconceived Badlands could address, the third will describe a way to launch a new network from within the IC, and the fourth will give a description of what that new network might look like.
What is not described here is the human infrastructure needed to design and implement these ideas. This is critical and will be the subject of subsequent posts. A final point for the introduction is that Badlands is a working title, but the actual branding of the new network, just like the design of the network itself, should be determined through market research.
Constraints
A marketer once remarked to me about DFINITY’s communications:
“Stop saying it’s super-advanced, and you have a great team! Everyone’s technology is advanced, everyone has a great team!”
The advice above is important to bear in mind for communications and marketing, but if we are talking about creating a Badlands-type network, it is essential to recognise the technical effort required to keep the IC running and moving forward. The DFINITY Foundation (henceforth DFINITY) is staffed by the best and has proven itself able to continue to make top-end hires. DFINITY is an incredible motor, flywheel, and drivetrain of technology development. This cannot be understated. DFINITY’s capabilities also need to be respected as a scarce resource. All efforts must respect the work done by the DFINITY team, lead to increased resourcing for DFINITY, and recognise its resources as finite.
Ongoing support of a Badlands network should not be placed on DFINITY. As such, there will be the need for external teams to provide maintenance and day-to-day operational support.
DFINITY technology really is the best of the best. It is and has been from the first the most visionary, elegant, performant, and truthfully the most viable base for realising the Web3 vision. Other technology suites exist:
-Ethereum/Solana/Cardarno/Hashgraph/Near/Polkadot etc. + Filecoin/Arweave etc.
-ZKnet/Starknet + name your blockchain/storage layer
-Holochain
-Urbit
-Secure scuttlebutt
But a challenging developer experience or complex infrastructure stack rules these out as contenders in the race for a new universal infrastructure platform — at least at this stage. For reasons the intended audience of this post should be well aware of, only the Internet Computer looks remotely plausible as a competitor to client/server-derived application hosting stacks. In addition, the IC offers a huge range of new capabilities that can only be found in cryptography-centric (read: blockchain-based) platforms. In Dom’s original Badlands post, he described a network made up of slower nodes to enable broader participation in node provisioning. However, reducing performance in service of a single end is drastic and limits the opportunities for the new network.
In designing a Badlands network, every effort should be taken to maintain or improve performance if possible. In pursuing opportunities, every option should be exhausted before resorting to a design that reduces performance. Another consideration is technological compatibility. The more the Badlands network diverges from the current IC, the more work will be required to repurpose development for one in service of the other.
The IC ecosystem is small. However, the community is dedicated and the quality of applications is spectacular. The community also possesses a healthy, positive culture that is an excellent germ with which to innoculate new members. The community and ecosystem must be preserved, nurtured, engaged, and leveraged. At no point should any action be taken that alienates or divides the community. Software composability must also be maintained at all costs. Fortunately, one important quality of the IC is that it can communicate using HTTP, even if the infrastructure base itself is non-contiguous, a dual network need not fracture the software ecosystem.
Opportunities
The IC platform had to be architected unilaterally. The Inverse is true for a Badlands-type iteration on the product: All ideas including those represented here must be validated through market research and community consultation before being accepted. This is an opportunity that the original IC could not take advantage of and is a key reason Badlands should be revisited.
This research should extend to all conceivable markets and user groups. This is how opportunities are really discovered and understood. A post like this one should be treated as a source of inspiration rather than a hard proposal.
Product over Platform
“The game will be won by product”
-Dom
The IC has always been marketed first and foremost as a high-performance decentralised hosting infrastructure. This marketing line has not gained the IC much traction. Badlands benefits from all the hard work DFINITY has already done as such, its focus should be on the applications that live on top of it. The most important of these is DeFi.
DeFi was the foundation of the last bull run and will be the foundation of the next. To benefit from this, Badlands should be devised with the goal of achieving the highest possible total value locked (TVL) across the ecosystem. It should launch with a token standard compatible with the external DeFi ecosystem and a complete suite of integrated DeFi primitives:
-Collateralised Loans
-AMMS
-Liquidity Mining
-Yield Farming
-Insurance
The value of DFINITY technology should be communicated by demonstrating successful products. As the underlying technology is already mature and populated by a number of DeFi teams, it’s just a matter of identifying gaps and filling them in.
Because DFINITY tech allows communication over HTTP it is not absolutely necessary for these DeFi primitives to be running on the Badlands network at launch. We just need easy-to-use front-ends deeply integrated with these services similar to Moritz Fuller’s waterslide.app frontend for Liquity.
Most people have a really hard time understanding DeFi terminology. The expectation for the future will be that they no longer have to. This ease of use will onboard more capital to the DeFi ecosystem and help catalyze the next bull market. Let’s get out in front of it. User interface design and application integration are where application development should be focused.
Apparent trends like NFTs, Metaverse, and GameFi could be seen as less important because they don’t necessarily involve financial leverage and locking of capital like AMMS and other DeFi primitives. However, they do represent excellent user on ramps. By sorting available applications in a kind of App Store, the most popular and usable can be surfaced for casual users, limiting the brain-work required to get involved.
Launching a new network with high-quality products in place will associate the Badlands brand with ease of use and the new age of crypto useability.
Tokenomics
If the Gini coefficient of token holdership grows over time, new stakeholders are prevented from making a proportionally meaningful investment in the token. This is described in traditional economies as R>G: R(rate of return on investment) is greater than G(economic growth). For the purposes of discussing tokenomics, we can limit our consideration to where newly minted tokens go.
There have been times this year when 8-year neurons were paying (R) over seven times the rate of new token minting (G). A very small handful of diehard Internet Computer believers are getting most new ICP and under the current scheme, their share of total ICP will grow over time. The effects of centralising token economics are already being felt in the IC. Creating an opportunity for new stakeholders to invest is one of the strongest reasons for a Badlands-type network.
Up and to the Right
The ICP graph slopes down and to the right. Investors look at this and fear a convergence with zero in the future. This not only makes the token less appealing to investors but negatively impacts the credibility of teams building on the IC when trying to raise money. By launching an incentivised Badlands with its own token in the depths of Crypto winter we can have a brand new graph that slopes up and to the right. It seems silly, but it’s actually really important to have an attractive price graph.
Making Money and Mining
The enthusiasm for the Badlands network was driven by the community’s desire to participate in the growth of the IC network. In service of this, Dom’s rendering of Badlands was cut-down in performance and could operate from nodes run in people’s homes. However, this rendering negatively impacts both the performance of the platform and the cross-compatibility of development for Badlands and the IC. If market research supports the need for a lower-performance network, it should be adopted. However, there are other strategies to facilitate network participation and decentralisation that can be tested with the user base before we sacrifice performance.
Ultimately people just want to make money from the growth of the network. This needn’t require a reduced hardware specification. Instead, an incentive of freshly minted tokens for running a node similar to how proof of work mining works may be enough to stimulate the emergence of node provider services and fractional ownership of nodes. It is also conceivable that running boundary nodes could offer a different flavour of mining for community members.
Crypto mining gained popularity in large part because it offered a way for everyday punters to generate an income while supporting their favourite network. This is a big opportunity the current IC has left on the table that Badlands could capitalise on.
Tokens and Governance
DFINITY holds a vast quantity of ICP and those tokens can be used for voting in governance. Yes, DFINITY chooses not to exercise this ability, but they have to be trusted not to do so. As a result, the Internet Computer is in a sense a trusted system. This undermines part of the purpose of decentralisation which (apart from resilience) is to obviate the need for trust. There are valid rebuttals to this line of reasoning but even if you accept them, valid rebuttals are not an effective shield from straw-manning propagandists who would use this to tarnish The IC’s reputation.
Right now Solana is coming under fire for exactly this issue but with Badlands, we have a chance to separate church and state, central bank and government, tokens and decision making. It’s an opportunity to remove the DFINITY token silo as well as adopt non-token-based governance.
Community Support
When DFINITY launched the IC, there was not the rich and enthusiastic community that has now rallied around the technology. As such the only user support had to come from inside the foundation. Badlands has the opportunity to launch with full community-driven user support from day one. This will have a positive effect on node providers, token holders, developers, entrepreneurs, institutions, and investors in projects building on the platform — everyone.
This can be activated through an incentivised, community-led support program. This program can extend beyond support for Badlands itself to the current IC and applications like the DeFi suite discussed earlier. We already have a community in place with over a year’s experience with DFINITY technology. This is a chance to more deeply leverage that community and create another way for community members to participate in and profit from the growth of the network.
Decentralised Funding
Currently, IC funding is controlled by DFINITY. A community-controlled ecosystem fund is a way to decentralise this. Why should DFINITY control the war chest? Why not give all undistributed tokens to a community fund for distribution via grants. These grants could fund development, support, and incentivise application development.
Decentralise Development
Grants from this fund could be used to stimulate the creation of at least one more protocol development organisation independent of DFINITY. This way two or more teams can develop DFINITY Technology. This is especially important to address the “key man”, or better put, “key organisation” risk. Additionally, this addresses one of our constraints: the scarcity of DFINITY development resources.
We can go beyond technical maintenance and use grants to fund new marketing strategies. These may be targeted at the audiences for both the Badlands platform and the products made available at launch. Plurality in both development and marketing is another opportunity Badlands presents.
Decentralise Marketing
On this point. Badlands can use an evidence-based approach to align marketing, messaging, and product positioning with these target audiences: What does the audience care about? What do they value? What do they dislike? Where do they see opportunity? How do they see the world? How do they see Crypto and where it’s going? How do they feel DFINITY technology can help them? What do they want? We can understand this and the culture of our target audiences afresh — before we reengage them with a new brand.
Products rarely speak for themselves (exceptions include the Mac, iPhone, and Furbee). We need to speak for our products in the languages and through the channels our target audiences use and ensure we align our brands in the same way.
New Faces
Badlands will be in a position to choose the people it selects as leaders and ambassadors. Rather than Dom acting as a lone figurehead, we can pick a number of public faces that can reflect and relate to our audiences. More than just an opportunity to mirror the image of our customers, having multiple people who can share the burden of public life and collaborate on messaging for multiple audiences will lead to less burnout, and higher performance, more prolific, and stronger engagement.
Execution
‘It has not yet been determined whether Badlands should be a separate sister blockchain that is interoperable with the Internet Computer (in which case, a BDL token genesis ledger would be created that replicates the ICP token ledger at the time of launch), or the Internet Computer should just support a new kind of “Badlands subnet blockchain”.’
Dom’s post describes two ways Badlands could be implemented. Many of the opportunities described above require a new token with new economics and a new governance system so this description will assume the creation of an entirely new network supporting a new token.
To simplify development and launch, rather than create this new network from scratch, however, we can deploy a new cluster of system canisters on the existing Internet Computer that node providers can configure their nodes to follow (These are the replica nodes. It has nothing to do with following in the NNS). This cluster of canisters will assign those nodes to a new hierarchy of subnets under those system canisters.
We can use this to set any rules or configuration preferences we wish: DeFi-optimised subnets, EVM subnets, private subnets, application-specific subnets — you name it. Nested subnets defined by different clusters of system canisters are a good way to diversify the features and run time environments of the IC. The catch is that the new network’s system canisters are always controlled by the overveiling network.
So how do we take these nested system canisters and subnets and give them the sovereignty of being self-hosted? One way we might imagine this working is to mirror the canisters on one of their own subnets and then gradually migrate the participating replica nodes over to following the self-hosted mirror. There are specifics related to re-starting global consensus that would need to be worked out but let’s handwave those issues away for today.
The next major infrastructure piece is the boundary nodes. Each of these is configured to query a system canister that acts like a phonebook — it holds a registry of all nodes, subnets, and canisters as well as their location in the network. When someone wishes to access an IC-hosted application, they enter a URL, which connects them to a boundary node and provides a canister ID to that boundary node. The boundary node then looks up the ID in the phonebook canister, finds the subnet that is hosting it, and relays traffic between the user and one or more nodes in that subnet.
Because the Badlands network will ultimately be separate from the current IC, it will need its own phonebook canister. This does not prevent Badlands boundary nodes from directing traffic to the current IC, however. All that is required is for the Badlands phonebook canister to also mirror the IC phonebook canister. In fact, it might be that Badlands is an opportunity to create an incentivised boundary node infrastructure for both networks. This would be especially sensible given the inevitable integration of software running on both networks. Given the likely future prevalence of high throughput applications like video hosting on Portal, a robust boundary node infrastructure is something we should be thinking about.
Configuration
The Badlands design and implementation should be the result of both a thorough policy creation process (this will be discussed in subsequent posts) and product development process. What follows is just a list of ideas.
Governance
There are several ways we could address governance. An easy one is to ignore tokens and say one replica node, one vote. This is basically the way proof of work does governance. We could keep the follow system to simplify participation. One benefit of this approach is filtering governance to represent only those actively participating in the maintenance of the network rather than passive token holders. This filter also has negative outcomes though — like creating a minimum capital investment required to have a say in governance. The appropriateness of this solution, like any other, needs to be determined by weighing it’s ability to address our stated opportunities while considering our constraints.
From a political perspective, by using a different filter to the IC, Badlands is controlled by a different culture. This is important to enable Exit. If a segment of the community is unhappy with the way the IC or Badlands is being governed, it doesn’t help for them to have an alternative solution governed by people who have been selected through an identical process.
The alternative must be sufficiently positively differentiated in both features and governance but still deeply interoperable to enable an individual’s drive for Exit to be satisfied without forcing them to leave the ecosystem.
Economics
It’s been said that good public policy is simple in design and has one clear goal. The more complex the design, the more that can go wrong. Trying to do too many things at once risks doing everything poorly rather than one thing well.
So what should be the goals of “Tokenomics”?
In a sentence: The goal of all token economics should be to support an effective monetary system to facilitate payment between a network of service providing (node provider) and consuming (application) actors, within the sovereignty of the network.
In a perfect world, this is where we would stop. There is another quality that lends itself to blockchain success, however, and that is having a large token holder base. This takes us back to the Gini coefficient of token distribution. But rather than try to design our economics to achieve a certain degree of token distribution, let’s start with something that works for our stated goal and then test the centralising and decentralising forces it brings into play.
- All tokens in the hands of the public as well as all undistributed rewards are mirrored on the Badlands network and fully liquid. This excludes DFINITY-held tokens which are not mirrored.
-10% of the token base is issued as seigniorage annually.
-These tokens are distributed to node providers with a cap of USD 10,000 value per node per month.
-The node providers allocate a proportion of rewards to the boundary node network via the governance system.
-If the USD 10,000 per month cap is reached, additional tokens go to a governance-controlled fund to support development.
-Tokens are converted to cycles that are burned to pay for network services.
If there are low barriers to entry for node providers, efficient pricing (this is a fantasy in today’s environment but let’s allow it), and growth in demand greater than seigniorage, price will increase over time. Node providers are forced to sell some of their earned tokens to cover hosting and capital costs. This forces the tokens into the market, decentralising their holdership. Dilution is offset by burning tokens for use.
A shallow assessment for positive token distribution trends yields a favourable result. The same forces that drive token distribution ensure market liquidity for those wishing to pay for network services. This scheme, at least on the surface, appears to satisfy our primary need for a functional monetary system and secondarily, our preference for a trend toward a lower Gini coefficient of token holdership.
Node Providing
Rather than enforcing KYC, the only barrier to connecting a replica node to the Badlands network should be a bond of $20,000 worth of Badlands tokens. If the node provider doesn’t wish to post a bond they can simply run a proof of work until they have earned it. Nodes without a bond are not placed in the cue to be added to the network.
Once the bond has been posted or earned, the node is placed in the cue and immediately begins dispensing rewards to the node provider. If the value of Badlands tokens drops such that their bond is no longer worth 20k, earnings cease to be dispenced and go to top up the bond. If the value of Badlands token increases, the total bond held over and above the current value requirement is made available for manual disbursal.
Upon electing to leave the network, a disconnection process relocates data and network functions from the node to other nodes in the network. Once this is complete the node can be disconnected and the bond released in full. This process should take a short amount of time and the node continues earning until the moment of disconnection.
Node misbehavior is determined by a set of pre-defined and algorithmically assessed criteria. Nodes are ranked by performance according to this criteria and the lower 0.1% of nodes lose part of their bond every week. If a node’s behavior is aberrant beyond a pre-defined threshold or disconnects, it loses part of its bond according to a three strikes per week rule, where the final strike leads to loss of all bond. This incentivises node providers to seek a diversity of robust hosting options for their nodes.
Boundary nodes provide a service to the network but are not as capital intensive to operate as replica nodes. There may be some possible attacks with boundary nodes but ignoring those for today, let’s allow anyone to run a boundary node. Boundary nodes could be incentivised on a per-relay basis, or all operational nodes could share equally in a boundary-node reward pool set aside from the 10% seigniorage by the replica node providers.
Conclusion
What is described here are some very simple tweaks that revolve around a basic principle: The primacy of the node provider. But does this system work? First, we should test it against our constraints and then consider its ability to address the identified opportunities.
Does this approach respect the DFINITY Foundation and create an avenue to promote and financially support their work? Yes, it does. There are minimal changes to the actual IC software beyond configuration changes and simplifications. It is very similar to the current IC in all respects meaning most work performed by DFINITY should be able to be adopted by Badlands. Also, by creating a governance-controlled fund it is possible for the Badlands governance system to offer grants to DFINITY, and respond to requests for funding. It also has the potential to support the existing IC with an incentivised boundary node infrastructure but makes no performance concessions. Finally, as was the driving motivation of the original Badlands post, it creates avenues for community members to participate in the growth of the network and have a greater say in governance than is currently possible while at the same time giving current token holders greater proportionate share of Badlands tokens than they currently have of the ICP token.
What about creating room to address our identified opportunities? This is more a question of flexibility than an assessment of the stated design. By being compatible with all existing software it is possible to launch Badlands as described here with all existing IC applications This would be best arranged through fresh UIs rather than competing implementations. The economics as described do pass a cursory Gini coefficient test and given all tokens will be liquid at launch, sell-side pressure should be significant enough to keep the price down while the network establishes itself. At the same time, the DeFi protocols Badlands launches with will enable price discovery and the creation of the price data we need to draw our nice upward sloping curve.
There is no built-in point of centralisation and the cap on replica node provider returns ensures token silos are not created due to minimal participation in the early days. The one token silo that is built into the system, the community fund, is under decentralised and potentially anonymous control. This fund can be used to support DFINITY, as described above, and it can also be used to support new protocol development and marketing teams. These are both opportunities for greater community involvement in the project.
It makes sense from a birds-eye view, however in real life, running hell-for-leather into launching a new network is likely to lead to more issues than we expected to solve. In planning an effort like the one described here we need to lean on relevant prior arts: product development, usability engineering, public policy making, political theory, economics, corporate strategy, marketing, qualitative/quantitative research & analysis, the list is as inexhaustible as the considerations are essential.
These considerations will be the subject of subsequent posts. I hope this has been as enjoyable and stimulating to read as it was to write. Crypto winter is tough but it always thaws. Now is the time for contemplation and making big plans. Think about where you want to be when next the bulls run. Hang in there everyone. Fallsy out.
