Do You Need to Cluster?

Think of cluster incentives as the opposite of the heat death of the universe.

Heat death is a theoretical end-state in which the universe has flat-lined. All energy is evenly distributed, every star has burned out, there are no hot spots. It’s all the same. Boring.

Cluster incentives serve the opposite goal, driving high value into specific areas.

Clustering in DePIN refers to groups of devices that provide more value at higher density than the sum of their individual contributions.

Instead of a network being perfectly spaced out around the globe including the thousands of places it’s not needed (say, a single Virtual Power Plant node on a hut in Antartica), cluster incentives drive high density node deployments that allow a DePIN to capture maximum value small-scale network effects as they grow.

Still, before we get too deep into what cluster incentives are, how to design them, and a few examples, we’ve got to ask:

Do you really need cluster incentives?

Your DePIN project needs to set up the rewards to create the network you want. If you need a more evenly spread out network, you probably don’t want clusters, and you’ll set your rewards to diminish once an area hits a certain density cap.

In a network requiring evenly spread out coverage like the Helium LoRaWAN network, the 100th LoRaWAN hotspot in Manhattan has no value, and placing 90 of those hotspots somewhere else would build a more valuable network.

If you need to hit a threshold of coverage in order to get a city (or region) up and running, you’ll want to use cluster incentives to increase rewards as an area hits a specified density, and diminish rewards for placements where density isn’t high enough.

A Virtual Power Plant (VPP) network needs cluster incentives. 100 homes spread evenly across America trying to join a VPP actually generates negative value for the project. VPPs need cluster density in order to operate, and paying out tokens for widespread solo deployments leaks value from the network.

The last conceptual piece to all of this (before we get into details) is redundancy. You’ll always want far more redundancy than a similar web2 project. You’re dealing with individuals here, not businesses, and the long term value of the data stream they provide typically isn’t enough to encourage any kind of heavy lift on maintenance.

Now we have the basic concept, let’s move on to when cluster incentives are appropriate, how to design them, and a few examples.

Studying incentives for clustering isn’t new; back in 2018 the Brookings Institution discussed aspects of clustering benefits. They were referring to clusters of enterprise (and really, humans), but the core concepts of the synergistic effects can be ported across to DePINs.

Technology Determines Cluster Size

How you design cluster incentives (and even whether or not you should use them) is first determined by the core technology of the project.

For example, GNSS is based on a general radius of about 20 km from any given base station. WiFi only provides coverage out to about 300′, and Virtual Power Plant systems consist of single buildings aggregated together across a defined region usually called an ISO (Independent Systems Operator). ISOs are huge; there’s one for California, one for Texas, one for the entire midwest, and one for New York.

Each has different requirements for cluster size and density, and different reasons for why they might use cluster incentives.

A cluster of GNSS base stations might need 4 inside of 1,800 square miles to provide coverage and redundancy, whereas a cluster of WiFi stations might be 6 inside of 20 acres for the same goal.

A Virtual Power Plant (VPP) cluster size might be 200 houses in some subsection of an ISO (although more accurately it’ll be some amount of megawatts, not houses). In that case, cluster incentives won’t be focused so much on redundancy and coverage as they are on gaining the critical mass needed to negotiate with the ISO.

As the US Dept of Energy has noted, “Utilizing the concept of VPPs allows generation and integrated utilities to treat them as another operational
plant, with standard attributes including maximum capacity, minimum capacity, ramp-up, ramp-down, etc.”

VPPs have the clearest fit need of cluster incentives. 200 hours split across 50 different states can not deliver the concentrated effects of energy from 200 houses in a single ISO.

Why Use Cluster Incentives?

With the first-principle limits and the demands of the technology in use as your guide, the next step will be to set clear goals for cluster size, density, and redundancy.

Well formed cluster incentives develop correct-density networks faster with less wasted capex, and without wasting tokens rewarding people who deploy in areas that are unlikely to hit whatever minimum threshold exists.

Cluster Incentive Examples

VPP cluster incentives are under development as we speak, but it’s useful to look at some of the groundwork other DePINs have already laid in this area.

Pollen approached clusters first with basic geographical boundaries, not allowing people to deploy outside of predetermined cities.  They enforced this in part by only shipping hardware to certain zip codes, then followed that with algorithmic rewards that encouraged clustered deployments and simulated demand.   

This combination of physical guardrails and math is a good example of what biologist E.O. Wilson calls consilience, “the ‘jumping together’ of knowledge by the linking of facts and fact-based theory across disciplines to create a common groundwork of explanation.”

As an aside, this is what Gold Hawks does, combining lessons learned from incentive design, community management, messaging, and overall strategy from a wide span of DePIN clients to develop custom ways for your DePIN to succeed.

Pollen’s system used gamification combined with inherent greed to get local deployers pooling resources to plan out deployments that maximized their local rewards, ultimately creating a cluster of deployment partners, albeit a malformed one.

Pollen obviously gave up before they ran out the experiment, although in the meantime we were able to observe and capture valuable lessons.

XNET has taken a similar approach with shipping to specific locations, then layering in gold (boosted) zones to fine tune which parts of the designated areas were even more important.  

Using this system, XNET got coverage in the cities they wanted, although they were stymied by a different problem of token price and token allocation.

The combination of a low token price, small allocation, and a narrow carrot/stick setup limited the development of the CBRS network, though that may seem prescient as they didn’t waste that many tokens before moving away from CBRS.

What Goes Into A Cluster Incentive?

Cluster incentives are generally made of three parts: Density, Size, and Redundancy.

Each part is determined entirely by the desired network outcome for the project. The central theme here is:

Set up your rewards/incentives for what you want your network to look like.

Simple in concept, this can be incredibly complex. A well-designed cluster incentive will take into account the technology deployed, the desired outcomes, time horizons for the project, and how long a deployed unit is likely to stay online. This last piece is a function of the robustness of the technology as well as the incentive itself.

To add in a real-world example to emphasize the need for extreme redundancy, despite being able to function for years, 70% of all Helium IoT Hotspots were not operating 2 years after being placed. Even with that loss, enough redundancy was built in that Helium still has the largest LoRaWAN fleet in the world with nearly all developed areas of the world covered.

A well designed cluster incentive will avoid “lean redundancy”.

With DePIN you’ll need many more nodes than you might think. If you’re going to err, err on the side of “more”.

Before moving on, we’ll leave you with a caution on cluster incentives.

Don’t copy/paste! You can’t just look at another project and duplicate their incentives. Every project is different, and even direct competitors will have dissimilar enough ecosystems that a “good” cluster incentive can’t just be ported over. Incentives need to be a part of the fabric of the whole.

We know it’s tempting. Incentive design is a lot of work, and it can seem like that “other” project’s incentives are close enough. It’s unlikely that they are, and the cost in tokens and breaking trust from getting your incentives wrong is enough to severely damage a project.

If you need help, bring in experts.

Balancing Precision & Speed: Trade-Offs in Clustering

Cluster incentives can be imprecise. You’re relying on the emergent properties of the interlocking incentives, not on the typical web2 method of “I want this thing in that exact place.”

This can be frustrating for our more precision-minded founders. Imprecision is traded for speed of deployment and “good-enough”, which in the early stages of DePINs is the critical element.

Boosting Cluster Growth: Incentives Beyond Technology

Clustering works best when combined with identifying a general target area (hex boosting, which we’ll cover in a future article), then designing incentives for clusters in that area.

Boosting can range from building in custom incentives for a geographical region to low hanging fruit like hosting events in targeted locations or sponsoring meetups led by community members.

Clusters may not get the specific location you want, but with enough clustering including the critical aspect of redundancy, you’ll bring forth a robust network across the target area. 

The Future of Cluster Incentives: Towards Fairness and Scalability

Properly designed, clustering incentives lead to local network effects, ideally combined across multiple locations to create both the density and network size needed for a successful project.

Cluster incentives are a promising solution for optimizing certain DePIN networks. Combined with targeted areas, they can create the crucial density a project needs to deliver the service coverage or data streams necessary to meet demand.

Well designed incentives require careful considering of the technology, density, time-horizons, and redundancy required for a healthy and value-generating cluster.

Given the complexity involved, it’s vital to approach clustering with a full quiver of expertise in incentive design, carefully thinking through to second and third order effects. After all, it’s no use to your project to have someone driving around town with 200 of your units in their car just because you missed a key incentive perversion. Yes, that happened.

While much of the math for incentives was developed as far back as the 1950’s, the modern day application of it to DePINs requires both fundamental understanding combined with skill in art that only comes from experience.

Max & I spend a fair amount of our time studying and understanding incentives, including the details of well formed cluster design. If you’d like our expertise applied to your project, don’t hesitate to reach out!

Disclaimer: Financial Interests and Consulting Services Disclosure

This blog post may contain references to various cryptocurrency projects, tokens, or assets. It is important to note that the authors of this blog post and Gold Hawks & Associates may have a financial interest in some of these projects or may provide consulting services to them, or in some cases, both.

The information and opinions expressed in this blog post are intended for informational purposes only and should not be construed as financial advice or a recommendation to invest in any specific cryptocurrency project. Readers are encouraged to conduct their own research and seek the advice of qualified financial professionals before making any investment decisions related to cryptocurrencies or any other financial assets.

The authors and Gold Hawks & Associates are committed to providing accurate and unbiased information, but it is essential to understand that our financial interests or consulting relationships with certain projects may influence the content presented. We aim to maintain transparency and integrity in our content, but readers should exercise due diligence and consider potential biases when interpreting the information provided.

Investing in cryptocurrencies involves inherent risks, and market conditions can change rapidly. It is crucial to make informed decisions and only invest funds that you can afford to lose. The authors and Gold Hawks & Associates assume no responsibility for any losses or damages resulting from actions taken based on the information presented in this blog post.

By reading and engaging with this content, you acknowledge and accept the potential conflicts of interest disclosed in this disclaimer.