Would recommend the following articles from UMA protocol before reading this article. Atleast read the first one.
A synthetic is an asset whose price is determined by a function f(x) where x = (x1,x2..) could contain any number of asset prices or real world data points. A synthetic is collateralised by an asset y. Note that x and y can be distinct.
For a synthetic to be secure, the collateral backing the synthetic must exceed the value of the synthetic for the lifetime of the synthetic. Synthetics like DAI have an eternal lifetime, whereas most synths currently on UMA have a defined expiry. Options too can be regarded as a form of synth.
Upper bound on synth price
The ultimate collateral for the synthetic provides an upper bound for the price of a synthetic. Suppose we create a synthetic whose value depends on BTC price, and it is backed by ETH. Suppose the synth is 3x collateralised. Minting $1000 of synthetic BTC requires $3000 of ETH. Now suppose BTC goes up 6x in price. The user who originally minted the synth is under no obligation to supply more collateral to maintain backing. No one else is obliged to do this either, unless they are provided a sufficient additional incentive. This means the user holding the synth must either liquidate their synth for the underlying collateral immediately, or continue holding the synth with the knowledge that it is no longer fully backed.
Collateral transformations (aka collateral liquidation)
Collateral can be transformed from one asset to another to ensure backing. For instance in the above example, there could exist an option to liquidate ETH collateral into WBTC collateral during extreme circumstances. WBTC collateral will naturally always be sufficient to back synthetic BTC. So under normal circumstances the collateral can be be a different asset than the ultimate collateral. MakerDAO too follows this - under normal circumstances their synthetic dollars are backed by ETH, whereas if ETH price drops significantly, ETH can be liquidated into USDC to ensure full backing for DAI.
The ability to transform collateral depends on available liquidity and ability to handle price shocks. Liquidation incentives must be sufficient to handle these.
Liquidity can be sourced from off-chain sources to ensure backing. Suppose WBTC doesn’t exist or have sufficient on-chain liquidity. Synthetic BTC can still survive due to the existence of off-chain liquidity for BTC-USD. A user can back synthetic BTC with collateral, sell it on-chain for a premium, and short the corresponding amount of BTC off-chain to get low-risk yield on his collateral asset. A premium could also be replaced with an interest rate that synthetic holders must pay over time. Since the traditional world has a high demand for yield on USD, it makes sense for USD to be used as collateral for such bridging operations. Such bridging also partly solves the earlier problems of there not being sufficiency incentive to put up more collateral to ensure backing.
Such bridging is immensely powerful, and democratises access to all forms of assets. Anonymous billionaires could access the liquidity of the US equity market or the Russian diamond trade. They just need to pay a premium to the other person on the trade bridging the liquidity. This other user too cannot easily be tracked. As far as their on-chain activities aren’t traceable, they are simply short certain assets.
It is hard to impose any form of regulation on what form of synthetics can be created or access them - and equally hard to regulate those bridging liquidity to them. This naturally bypasses all the regulations currently present. For instance, private equity imposes considerable restrictions on who can own it or how it can be bought and sold. Such liquidity bridging can create liquid publicly-listed unregulated markets for private equity from day one, provided the liquidity brigder is able to hedge their exposure while following regulations.
“Shady” markets too can be invested in, with the bridging users alone having to deal with the regulations and other risks in investing in that market. Typically the harder a market is to access, the higher the premium that may be earnable by bridging liquidity for it.
One can also create entirely new forms of assets and risk transfers. A major activity of current financial markets is risk transfer - one could directly create new and more innovative forms of assets that enable this risk to be transferred across the globe.
Impact on underlying assets
One tends to assume that the price of derivatives and synthetics is independent of the underlying backing assets. Consider an ETH call option backed by ETH. Demand and supply for ETH affect ETH price, and demand and supply for the call option affect the option price. Since the ETH call option price depends on the price of ETH, demand/supply for ETH impacts demand/supply for the option. But if there is sufficient demand/supply imbalance in the option, the demand/supply of ETH itself can be affected.
An anonymous entity buying synthetic Google stock on ethereum could in fact move up Google stock prices given enough capital.
This impact is especially important on underlyings that are not liquid or resistant to movement. For instance, UMA has launched synths whose price depends on the TVL of a given protocol. TVL however is easily manipulable. Adding $1 billion of TVL to Aave may cost only $5 million worth of risk for the short period that TVL must exist in Aave to win the bet defined by the synthetic. Therefore one must take care to pick those underlyings that are proportionally liquid and hard to manipulate as compared to the synthetic. Assassination futures, commonly discussed either jokingly or as an example of an unethical market, too face the same issue. The bets on the synthetic or future may be arbitrarily large, whereas the cost of impacting the underlying (performing the assassination) is a constant that does not scale.
Synthetic pay and control
The underlying being used for synthetics needn’t be an asset, as you may have already seen. For instance you could define a synthetic that evaluates to $1, if a certain ship completes a certain route at a certain schedule, and $0.2 if not. This allows payments for real world jobs to be paid on-chain. If the ship operator buys this asset for $0.5, they have both a positive motivation (payment) to complete the job and a penalty for failure to complete the job. Such carrot-and-stick is synthetic control. Synthetics can be made arbitrarily complicated to replicate the complex forms of contracts, trust and incentives present in the real world.
In theory, entire corporations could be synthetically controlled by anonymous entities. The hardest job ofcourse, is getting reliable oracles to report these events. This likely requires public or semi-public information (zkproofs, anyone?), it also requires economic guarantees to hold. UMA has begun exploration of these economic guarantees. Existing trusted entities can also be used to bootstrap trust, this has been further discussed in my article on the oracle problem. Trust is a complex blend of economic guarantees across repeated games mixed with psychological factors such as morals and reputation.
It is hoped that this deep experimentation into new forms of assets could also solve collective action problems such as say, climate change. Carbon emissions and weather futures are already being traded, however access is restricted, and their trading is still somewhat complex. Synthetics could be create to open access and repackage these assets in forms that are easily understandable by the general public - so that they can then invest in these and impact change in the underlying events.
Creating weather futures does not entirely solve the collective action problem - since private sacrifice is still leading to public good. However this can be overlayed with a social movement that creates social pressures and motivations to engage. Social pressures and motivations are one of the strongest forces for change, since our desire for social appreciation is it times even more fundamental than our desire for financial gain. It is possible that democratising these synthetics will make it easier for one to layer social movements on top of them.
DAO-like mechanisms may also bring us closer to solving collective action problem by laying out financial incentives to cooperating and disincentives for defecting.