I prefer to begin with this based on the fact that most of the debates on oracles start at the wrong point. They tend to begin by stating a big claim of oracles being the foundation of DeFi. That is true but evasive of the real problem. The thing with it is that most on chain systems fail silently. They fail quietly. They fail due to the fault that the data feeding them is misplaced, or lowly created or incorrectly created. And when it does occur the damage contagions fly.
I have witnessed this on several occasions than I would prefer to acknowledge. Any person trading perpetuals involved with chain options or more advanced derivatives of DeFi has felt this way. Price glitching does not go gracefully. Liquidations fire. Vaults drain. Jobs that are supposed to last are lost. The anarchy proceeds without adventure or slauder. Just bad data.
During a sharp move in BTC I in the recent past, a small derivatives protocol had to freeze. Not that there was a dearth of liquidity. Not because users panicked. However, as their oracle took a few seconds. That small delay was enough. Healthy positions were wiped off. There was no attacker. Just timing failure.
It is precisely the experience that made APRO attract my attention.
Not that it is expected to be faster. It is not more decentralized as it claims. Everyone claims that. The interesting point is that APRO approached the oracle problem as an engineering failure and not as a marketing opportunity.
This article is not hype. It is a realistic examination of why oracles continue to break the manner in which data actually traverses chain and why APRO is attempting to resolve the issue in a more sensible manner.
It is the section that nobody likes to discuss. Blockchains do not want data. They tolerate data. On chain systems are deterministic in nature. External data is messy. There is an agreement of disagreement between prices move APIs. And still DeFi protocols tend to assume that one number being pushed on chain is the absolute truth.
My own experience is that the worst oracle failures are not necessarily as a result of direct attacks. They are based on edge cases which no one expected.
Liquidity assets that go up abruptly. Off market hours of a stock or commodity. This is the price of floor during thin volume NFTs. Data within the gaming can be spoofed. The random that appears to be random until it reveals patterns.
I once tried a GameFi which the reward was based on the randomness of an oracle fed. All was well until one of the cluster clusters of validators began to predict. The game economy failed practically overnight.
Therefore now when I consider an oracle I do not question whether it is decentralized enough. I question what it is like in times of wrong?
APRO seems to have been created keeping that question in mind.
APRO is fundamentally a decentralized oracle system and it provides off chain data to on chain applications. That is a simplistic description but the design decisions beneath it are significant. APRO is able to support data push and data pull models. It gives support to various asset types. It uses a two layer network. And it provides AI powered verification in a more sensible than marketing manner.
Majority of oracle systems compel developers to one direction. APRO does not. and that is a greater flexibility than it may appear.
Firstly, we will discuss data push. This is the model that most individuals have in mind. The prices and metrics are constantly pushed at a chain either on a fixed time basis or by hitting a threshold. This is necessary to liquidation engines of perpetual futures lending protocols and automated market makers.
I have been involved in trading during volatile periods where seconds of lapse cost a lot in terms of profitability and resulting forced exit. During such times push based feeds cannot be optional.
There is APRO push system that is configured to be responsive on real time, only that it has extra layers of verification. This is important since speed that is not validated is only quicker failure. The bad data is more lethal than the slow data during the high volatility periods.
It is now that data pull involves APRO doing something that is underestimated by many people. Not all the applications require regular updates. Others just require information at particular points.
Options settlement. Insurance payouts. Event based triggers. Historical snapshots. Custom verification.
It is both unproductive and unsafe in the instances of data pushing that is continuous. APRO data pull model enables the smart contracts to only request data when necessary. This saves the consumption of gases minimizes noise and reduces unnecessary updates.
This is very valuable as far as a builder is concerned. You are not paying something that you are not in need of. You do not flood the chain with useless notifications.
The other notable thing about APRO design is that it has a two layer network. It is one of those ideas that sound as self-evident but are hardly implemented properly.
APRO divides the workload into both an off chain layer and an on chain layer. The off chain layer works with data aggregation processing and AI driven checks. Verification consensus and final delivery is done on the on chain layer.
This division is crucial since it is costly and time-consuming to put everything on chain. Moving everything off chain is not secure. APRO balances the two.
In the old type of finance, raw market data are not cleaned in the same system which does the trades. You separate concerns. APRO mirrors that structure.
AI powered verification layer is yet another field where skepticism is healthy. AI is a frequently used term in crypto as a buzzword. In the majority of cases it is not a lot.
APRO is also not a decision maker but a filter. The AI layer assists in identifying anomalies to recognize outliers and point out suspicious behavior and minimize false positives.
Think of it as a sanity check. Is not a substitute of consensus.
I have observed oracle systems with a single bad source making the average slightly skewed and the skew makes a huge downstream impact. The APRO approach mitigates such a risk by doubting data prior to it making it into the chain.
It is not perfect. No system is. But it is obviously much better than blind aggregation.
Another field on which many oracles silently fail is randomness. Any person who has constructed a lottery game NFT mint or randomized reward system will understand how difficult it is to achieve true randomness on chain.
Pseudo randomness is foreseeable. Randomness that is not on-chain takes trust. Chain randomness of chain is limited.
Verifiable randomness is among the attributes of APRO. This is important. In one of the audits I conducted previously, block producers could affect randomness in a protocol. No one was paying attention until someone started payouts becoming suspicious.
APRO randomness design emphasizes on verifiability and provability. Outcomes can be checked. Manipulation is noticed. There is a reduction in the assumptions of trust.
In gaming DAOs and fair distributions, price feeds are just as important as this.
Another field that APRO thinking is long term is asset coverage. APRO does not just support crypto assets, but also stocks real estate gaming assets and other real world data.
This initially appears as a feature checklist. However, it is important whether DeFi will be more than just a speculation.
I have witnessed tokenized real estate projects fail not due to regulation but due to poor price feeds. Valuations lagged reality. Liquidations made no sense.
The various classes of assets do not act the same way. They are periodically updated with various frequencies. They need various validation rules. That appears to be the complexity of which APRO is constructed.
There is also APRO in support of over fifty blockchain networks. The value is not so critical as the way integration is managed.
As I have observed APRO specializes in the areas of lightweight integration flexible APIs and compatibility with various execution models. It is important when being deployed across chains which are very diverse.
Another silent killer is the oracle costs. Large amounts of gas used during frequent updates and unneeded data pushes are all slow to run protocols. Not in a dramatic manner but in a gradual way.
APRO hybrid model assists in reducing unnecessary updates on chain computation and unnecessary fees. In the case of smaller teams this can spell out life and death of a product.
The questions are pragmatic as a builder. Is it possible to decide how data changes? Can I customize feeds. Can I verify sources. Is it possible to cut expenses when the activity is low.
APRO appears to be created to respond to those queries in the affirmative.
Naturally APRO will still have to prove itself. No protocol is complete. It should be able to endure black swan events. Incentives be it validators should stay in line. Adoption should expand to non-niche applications. Actual stress will be exposing.
I have witnessed excellent technology fail because of bad incentives. I have observed mediocre technology to be successful since it was shipped at high reliability. APRO architecture provides it with an opportunity. The rest will be determined by implementation.
On a personal level I think the most significant DeFi failures in the future will be data assumptions and not smart contract bugs. Assuming prices are fair. It is random to assume that it is random. Assuming feeds are timely.
APRO contradicts those assumptions by introducing flexibility and realism of verification.
It is not often that infrastructure projects are praised. Whenever things go wrong, they are held in account. The category that APRO is developing in is such that winning is dull and losing is devastating.
Provided APRO keeps its attention on the data quality, rather than on the hype verification at any costs and speed, and on the needs rather than on the narratives it has a fair opportunity to become a foundation.
Foundations do not trend. They are merely the supports of everything.
It is there that value is usually generated in reality.
