It’s no secret that building large-scale, fully decentralized applications is a challenge, but it turns out the hurdles have to do with more than just scaling.
Decentralized application or “dapp” developers frequently hit roadblocks, since ethereum – the go-to platform so far in dapps’ short history – can process only around 25 transactions per second, and the more transactions the network is asked to handle, the more each one costs the user. These limitations on transaction throughput are commonly referred to as “scaling” limitations, and everyone from the casual dapp user to ethereum founder Vitalik Buterin is keenly aware of them.
So when a particular dapp turns out to be less-than-totally decentralized, with parts of the software running on centralized servers, say, the solution seems obvious – speed up transaction throughput and reduce costs, and pure, blissful decentralization will naturally follow.
Turns out, though, things aren’t that simple.
Some of the most popular dapps that currently live on ethereum – which right now fall into two categories, games and exchanges – often retain centralized features, but the reasons have little to do with throughput and instead, revolve around user experience.
Take games – for developers to be able to make updates to a blockchain-based game, they typically put backdoors into the smart contracts.
Otherwise, said James Duffy, a co-founder of Loom Network, which develops ethereum-based dapps including a Q&A; site called DelegateCall, developers would only be able to deploy their dapp once and never be able to modify it.
“Obviously if you’re a player in the game you want the developers to be able to update it. You want them to be able to fix bugs, add new levels, add new features,” Duffy said.
On the other hand, decentralized exchanges (DEX) keep some centralization in their processes as it relates to their order books. The reasons for this approach have to do partly with security and partly with the difficulty of assembling a single, reliable order book across a large, distributed network of computers.
And while most of these dapp projects aim to decentralize further in the future, for now, they’re happy to work slowly through that process so that users have the best experience and don’t lose money.
Duffy told :
“No one’s ever built a complicated app and then launched it and it just worked perfectly on day one.”
The back door dilemma
CryptoKitties was the first decentralized application to gain widespread attention and a significant userbase.
It didn’t take long, however, for critics to notice that the game is not as decentralized as it initially seemed. Another Loom Network co-founder, Luke Zhang, wrote about the backdoors in the CryptoKitties code, which allows the company behind the game to pause it entirely or alter its closed-source breeding algorithm.
While the CryptoKitties team contends that keeping the breeding algorithm a secret makes the game more fun, this choice does mean users have to trust the company itself not to tweak the algorithm in a way that would undermine the market’s pricing of kitties (some of which are very rare and in turn very expensive).
Another piece of the Cryptokitties game that was under the company’s control, until very recently, was the art assets.
Without these, a player would still own that kitty they paid 250 ether for, but instead of being able to admire its green eyes and Himalayan, orange soda-colored fur, they’d have to admire the number sequence encoded in the non-fungible token that’s at the core of the game: 99ac5586a447g9gg44665775ddf71444488773384ccccdffc.
But according to Duffy, without this centralized control of the art, developers and players might abuse the privilege.
“What happens if someone uploads something illegal, like child pornography or something?” he said. “The nodes would have to have a way of censoring that to remove that data, or it would just be a complete free-for-all.”
And yet, CryptoKitties seems to be ready to take somewhat of a chance on this – on June 26 the company, now known as Axion Zen, announced that it had updated its terms of service, making it possible for third-party applications to use CryptoKitties art. And not only that but it was open-sourcing the ownership rights of the non-fungible tokens.
Still, it’s a far cry from complete decentralization, and Duffy acknowledged that there are potential pitfalls in CryptoKitties’ approach to its business model. But he said that launching a semi-centralized dapp and decentralizing it over time is “pragmatic.”
Kyle Samani, the managing partner of Multicoin Capital, echoed Duffy’s statements, saying, “Decentralization is generally a spectrum.” He called criticism of CryptoKitties’ centralized aspects “nit-picky.”
Duffy held out hope for more complete decentralization, though.
For instance, he argues that Loom Network’s approach – building a dedicated, scalable sidechain for each dapp and pegging that chain to ethereum – would enable full decentralization without having to settle for fossilized games that never add new levels or features. Updates could be made through hard forks, assuming players could come to a collective agreement.
Making a prediction, Duffy said:
“We’re going to see actual, real, fun games that are normal games that people want to play, except that they’re actually owning the in-game assets and they can do this on a fully decentralized platform.”
Order book blues
Centralized exchanges have been cryptocurrencies’ Achilles’ heel since the early days of bitcoin – MtGox being the most famous example of what can go wrong when transparent, decentralized ledgers meet opaque, centralized intermediaries.
For this reason, advocates of decentralization have long tried to build distributed alternatives. Examples on ethereum include Idex and ForkDelta, which according to DappRadar are the first- and second-most trafficked dapps over the past 24 hours.
The only problem is that both of these exchanges use centralized order books, as do most of their peers, according to Taariq Lewis, a veteran cryptocurrency developer who is building DEX technology – codenamed Lyra Protocols for now.
These centralized order books, which collect “bids” (prices offered by buyers) and “asks” (prices offered by sellers) to facilitate trades, are the norm in traditional markets.
Despite strict regulation, however, shenanigans abound on traditional exchanges. Spoofing, front-running and layering are just a few of the (illegal) tricks traders use to take advantage of each other, and this behavior is prevalent – if not worse – on crypto exchanges.
“Unregulated centralized order books are manipulation havens,” said Lewis.
A person familiar with DEX operations, who asked not to be named, added that some exchanges see order book decentralization as a way to avoid regulatory interventions. Exchanges operating centralized order books must either register as Alternative Trading Systems or avoid listing securities – which, as is becoming increasingly clear, many crypto tokens are.
Decentralizing order books, however, is anything but simple. Traders need a main order book that is visible to everyone, Lewis said, and ensuring that everyone sees the same bids and asks without a central intermediary is “an unsolved problem.”
He continued, “It’s not trivial. A lot of these things folks have been working on for decades, well before blockchain.”
Adding to these technical difficulties, Lewis said, decentralized order books can be easy targets for Sybil attacks – whereby one user or group of users creates hundreds if not thousands of identities in an effort to spam the network with information.
Still, exchanges such as Idex and ForkDelta have said they plan to decentralize their order books when the technology enables them to.
Lewis did not want to reveal too much about Lyra Protocols, but said the project was “looking into” attack-resistant, distributed order books. And Duffy said that a number of DEXs have reached out to Loom Network wanting to take advantage of dedicated sidechains for the purpose of decentralizing more of their processes.
As such, Duffy expressed confidence, saying:
“Give it a couple years and I’d say it’ll definitely be the case, just because it’s possible and if users demand it, then someone’s going to fill that demand.”