GitHub

Whitepaper

An Overview of GainForest

"Protect. Restore. Fund." - Greta Thunberg

The problem

Protecting and restoring our natural world is a resource-intensive task and conservation projects around the world depend on sustainable funding. Unfortunately, traditional donations require mutual trust between all parties. Donors must trust that communities are using their received donations for the agreed purposes. Communities, on the other hand, have to incentivize donors to donate to them. This usually happens through time-intensive and costly actions such as public marketing, scientific studies, expensive data collection, or third-party certification to verify impact. In addition, a middleman (such as a bank) usually sits in between the transaction of donors and communities, introducing significant legal (e.g. blocking transfers) and economic friction (e.g. large fees).

Our design goal

GainForest's design goal instead is to enable frictionless transactions by leveraging trust-enhancing technologies. We replace the need for third parties by introducing a self-enforcing smart contract that parks donation funds from individual donors. These funds are addressed to specific communities and conditioned to verifiable milestones. The smart contract is designed to channel funds to communities whenever milestones have been verified and to incentivize donors to contribute to causes they care about. Transactions are transparent and immutable thus replacing mutual trust between parties with automated and auditable proofs. Market mechanisms on top of the fund turn traditional one-time donations into continuous sustainable payments, and donors into investors.

Figure 5: Overview of all stakeholders and modules involved in the GainForest platform. Transactions either raise funds (black) or are token-based incentives (purple)

Our core architecture

GainForest's core architecture consists of three independent smart contract-powered modules that manage all transactions between investors and local communities: Fund, Markets, and Oracles.

  • Fund is the main component of GainForest's payment system which manages and channels funding from investors to communities. Investors park payments in the decentralized green fund contract and link it to specific communities. These payments can only be received by the community when certain milestones are fulfilled. The GainForest Fund generates interest over time and keeps an immutable log on the Blockchain while being completely self-enforcing. We will dive deeper into the implementation details in Chapter 3.
  • Markets are the building blocks of GainForest's incentive system. Markets are unique interfaces to develop and deploy decentralized environmental cryptoeconomics to enable novel incentives for engaging communities and private citizens. They leverage mechanism design to ensure optimal sustainable funding. The GainForest association develops and maintains our own open-source token-based funding model called Dynamic NFTs (or NFTrees), a virtual digital asset, to further connect investors and communities with each other. We will provide an overview of Dynamic NFTs and possible markets and games in Chapter 4.
  • Oracles are at the heart of GainForest's Proof-of-Care system. They are machine learning-based impact evaluators that leverage satellite and drone imagery as well as field data to detect and recognize ecological changes. If milestones are achieved, oracles send key tokens that link to verification reports. Communities receive these key tokens and can use them to unlock payments from the decentralized fund. This allows investors to exactly trace back the impact of their specific payments. The GainForest association develops and maintains our own open-source impact verifiers on top of Sentinel-2 data that are provided to communities. We will dive into more details in Chapter 5.

Did you know?

ESA's Sentinel-2 mission is providing public satellite imagery of the earth with a resolution of 10m per image pixel for every five days.

Figure 6: Inferred vegetation height of a machine learning-based model from remote sensing data in Switzerland (1m/px).

Previous
1️⃣ Our Vision