Ripple is probably one of the most polarizing coins in the crypto-ecosystem. I have been following the evolution of the Ripple project since shortly after it got founded in 2012, and I would like to describe it from a personal perspective, which is certainly an external one and not necessarily expert.
Ripple is a blockchain-based system that features a native coin, XRP, and token classes (which can have multiple issuers). It was founded by a company, OpenCoin, later Ripple Labs, which centralizes its issuance, development and administration.
The original Ripple
Ripple originated as a project of electronic money based on personal credit, developed by Ryan Fugger. Users extended each other trust lines up to a certain value, and then credit would be transferable between any two users in the network by automatically rebalancing trust-lines' states, an action named "to ripple". Such system does not need global consensus, nor a native coin. However, it requires that trust-line paths are online to be updated, which does not scale. In consequence, Ripple was re-founded by OpenCoin with a consensus algorithm between servers and with XRP as a token for spam prevention. Some XRP got initially distributed on the BitcoinTalk forum.
Ripple's history has two different episodes. In the early one, they start in the bitcoiners' side; they want to build a cryptocurrency system that is efficient and minimal-trust based, but still entirely decentralized and permissionless. In the later one, they fully embrace being a bank-chain (blockchain technology for banks) and trying to build a substitute for SWIFT: xCurrent, which does not use the network nor the XRP token.
The transition between those two eras is marked most decisively by two events. The first one, founder Jed McCaleb left Ripple to found a competitor: Stellar. He forked the code and started stressing the network for decentralization, and that quickly made it break, thus giving empirical evidence that Ripple's consensus method (which was, principally, McCaleb's invention) was not safe. By extension, that event evidenced that Ripple would not work in a decentralized setting, and indeed Ripple Labs axed their decentralization project. The second one, FinCEN determined that Ripple Labs was subject to its regulations, fined them, and forced them to stop offering the end-customer facing frontend (a webwallet known as Ripple Trade). Ripple Labs reached an agreement with FinCEN in which they would deploy an anti-money-laundering system on the protocol.
This ended all dreams of building a permissionless system. Soon after, they deployed also permissioned accounts, asset-freeze functionality (on the sole request of the issuer), and then moved to work almost exclusively for banks.
Liberty Reserve proved that government regulations cannot authorize a private electronic-money system that is simultaneously centralized and permissionless. As Ripple failed to decentralize, their only escape from that existential threat was to adopt a user-hostile, government-friendly position. Rather surprisingly, this has been incredibly successful, both for the company, Ripple Labs, as well as for the coin, XRP. Technological fail, business success.
But banks are adopting Ripple, right?
For banks, Ripple Labs has developed the Interledger Protocol, which is a messaging system with standard calls supposed to be easily integrable in the banks' backends, as well as other types of ledgers. This is marketed as xCurrent and it is made as a candidate that could replace SWIFT. It is a product fully different from the Ripple network, in which there is no intervention of XRP nor asset-tokens. Banks do not necessarily trust much each other: They may tolerate having accounts at correspondent banks, but do not want to hold other banks' IOUs, nor volatile tokens.
The role of XRP
XRP is occasionally defined as a cryptocurrency. However, even Ripple Labs dislike calling it like that, as much as it is indeed supposed to work as a currency and be backed by cryptographic tools.
Banks are not using XRP in any way. But it does not matter, as it is mostly marketing what backs XRP value. Big bank partnerships for Ripple Labs are interpreted as a support for XRP, even when it is for xCurrent and not the Ripple network itself. Still, on the Ripple network, financial institutions can issue their assets as they want. In that case, they retain full control, can authorize which accounts of which users are authorized to hold them, and can freeze them whenever they want. Despite those assets are often called IOUs, they work quite differently from real-world IOUs, which are under control of the secondary market trading them, not directly of the issuer as in Ripple. This property compromises the fungibility of those assets, which are thus even more fragile than Tehter-USDT.
The role of XRP in the Ripple project changed after FinCEN's $700,000 fine. Before that, Ripple Labs used to distribute XRP in significant quantities, both for free, as a reward, or as direct sale to individual investors. In particular, FinCEN found that a sale of XRP to Roger Ver violated the Bank Secrecy Act. After that, they have distributed exclusively to selected corporate partners and in much smaller amounts. As a consequence, a majority of XRP is still under control of Ripple Labs, even five years after launch.
Spam prevention From a technical point of view, XRP has always been a useless token, in a certain way. However, when the system was supposed to get eventually decentralized, XRP got marketed as a possible spam prevention mechanism. A decentralized network needs incentives to keep spam at bay, but for this it is only necessary that XRP holds any value, not necessarily a big one. On the other hand, in a centralized network, there exist more traditional spam prevention mechanisms, so the presence of XRP is not even needed.
Bridge currency Some have argued that XRP could also be used as a bridge currency. As the native token of the system, all accounts can hold XRP, and this could be leveraged as a bridging point for all the other assets used in the network. Inter-asset exchanges using Ripple's internal market system would exchange through XRP, which would thus need to be valuable. However, liquidity in this internal market is and has always been very poor and practically unusable. XRP huge volatility does not help either. Overall, the system could work well too with multiple bridges, which could be, e.g., USD of major banks; it is highly dubious that XRP could play a vital role there.
Use to transfer funds between exchanges Some users have mentioned that they use XRP for inter-exchange transfer. It is quick and relatively cheap, except on exchanges that generate a new address for XRP (and absorb the first 20 XRP as a reserve). However, XRP is not always used as base currency, so when a user wants to transfer her assets by exchanging them to XRP, she will often lose more in exchange fees and volatility risk as she would by transferring bitcoins directly. Typically, unless the bitcoin network fee is more than 0.4% the amount, it is not worth it to pass through XRP.
On the other hand, there exist recently released blockchain products like Liquid by Blockstream, which are specifically marketed as a way for cheap and trustless inter-exchange transfers of bitcoins and other tokens.
Using Ripple today
Today's situation is nothing short of a joke for an end user of the Ripple network: You need around 100 XRP to meaningfully participate in the Ripple network in any practical way (this is the bare minimum of 20 XRP + 5 XRP for every subsequent trustline and exchange order). This is $70 at today's value, so it is more expensive than a typical trading account, and certainly unsuitable for "the World's unbanked" as it was advertised years ago.
Those 100 XRP get indefinitely locked (with the bare reserve of 20 XRP lost forever). Then, there is up to 0.012 XRP charged as fee for every operation. That is usually cheaper than transfers of decentralized cryptocurrencies, but still too much for doing any high-frequency trading in the internal market; liquidity is poor in consequence and even the most liquid assets have spreads of more than 1%. These limits have not moved since c.2013, so it does not seem that there are any plans to adjust these amounts to the price of XRP.
Ripple exchange market: One of the most liquid assets are USD issued by Bitstamp (a big crypto-currency exchange). The issuer, not Ripple, charges a 0.2% per trade, even though trades are done outside of Bitstamp
What about smart contracts?
Ripple was also meant to become a platform for cheap and open smart contracts. Their initial design looked better than Ethereum back then. It was based on Google's NaCl (native client); they wanted something efficient, not reinvent the wheel, as Vitalik did with Ethereum. Unfortunately, then FinCEN fine come and that project evaporated.
Last straw: rippling disabled
Ripple's chainstate is account based. Each account needs to pre-reserve a space in memory for existence, and then for each balance and nonce. By default, all acounts have XRP, and then for every new asset balance, account owner needs to reserve a new entry that specifies it; this is called trustline.
There is a vestigial inheritance from the old Ripple (Ryan Fugger's) for which the assets where supposed to be used as personal credit. This is why trustlines specify an amount limit, even if that is no longer strictly enforced today. In that old Ripple model, lines of trust (trustlines) between different users would rebalance each other according to the limits. This action was called rippling.
However, the incentives of that got altered when trustlines began to be used by gateways (financial institutions that keep the issued balance in custody). Trustlines no longer represented trust in a direct way, but still gateways required them to function. Not all gateways where equal (some got hacked, some had high fees) and people started abusing that rebalancing system for personal gain. Abuse got so intense that the feature ended up disabled by default. There was an attempt to solve the issues by customizable settings to specify "gateway quality" parameter, but they did not work in practice.
Rippling, the very feature that gave its name to the network, vanished.
It is freezable, arbitrarily editable, and not even fair
Another bad consequence of FinCEN ruling was that asset issuers (gateways, most notably) got the right to arbitrarily freeze users' balances. This was a retroactive alteration of the issuance conditions, as assets issued and sold to users as unfreezable, and thus fungible, suddenly became freezable. This single alteration of the protocol changed many gateways' private contracts and terms of service.
Other than that, freezability is arbitrary, because it is done at the sole discretion of the issuer, and unlike many real-world assets that can be frozen only at the request of a court. It opened the way to selective scamming by shady gateways.
The internal exchange market is gameable. The ordering of trades is based on their hash identifiers, and trading orders specify a limit; this way, an exploiter can force them to execute at the worst rate and pocket the difference, as evidenced here. That exploit got somewhat mitigated, but it is fundamentally unsolvable.
Fail at decentralizing, fail at resist regulatory capture, fail as a smart-contract platform, fail of the rippling function, fail as a fair exchange system and fail at being cheap and attractive for end users. However, all those things above said, Ripple has been an incredibly enormous success in the way that the exchange value of XRP has skyrocketed. Ripple's many technological fails do not seem to have much impact on the value of XRP; for traders, this is what matters in the end.
Ripple is excellent at marketing and incredibly well resourced. It is also well connected with major banks and with American regulators. Ripple is also unique in the sense that it represents one of the few, if not the only, option in which end-user crypto-investors can have a stake in a bank-chain (blockchain for banks). This is because Ripple's competitors in this space (e.g., Corda or Hyperledger) do not have a native token, or it is not available to public ownership.
Partnerships between major banks and Ripple Labs, even if they are for xCurrent, mean that Ripple Labs is successful. Some insignificant testing of "Ripple network" technology, even if not using XRP at all, implies development support for it and, indirectly, for XRP. Banks can ensure healthy and maintained codebase later used in the open network and thus hopes for the token.
Chart of the historical price of Ripple in bitcoins; Ripple has become bigger than Bitcoin by total market capitalization when it has exchanged at a rate above the red line.
By total market capitalization, Ripple has already been bigger than Bitcoin multiple times. Ripple does not have (significant) real-world use cases yet, so any speculation on the value of XRP is entirely founded on future uses; today's circulating supply is thus not relevant to its valuation. This is justified valuing it by total capitalization instead of by circulating capitalization.
I disclose owning a quantity of XRP, as well as assets on the Ripple network.