Report: The EOS dapp ecosystem is mostly bots
The exchanges, casinos, games and digital ecosystem are increasingly filled with competing bots.
A new report from AnChain has used AI to identify bots – because it takes one to know one.
According to the report, more than 50% of EOS accounts and 75% of EOS transactions are bots. The primary purposes of these bots, AnChain suggests, is to boost dapp rankings by simulating the appearance of real use, to wash trade dapp tokens and to automatically scrape up the various incentives dapp creators offer users.
While other networks are likely also susceptible to the same sort of fakery, the report says EOS is probably especially vulnerable because transactions are free for users. Rather, the cost is shouldered by developers who buy network resources on the EOS RAM markets.
And while you'd generally expect human users to make more transactions on networks with zero fees, and bots to still make some transactions on networks with fees, it's probably no coincidence that the zero fee networks of EOS and TRON have such a disproportionately high transaction count compared to Ethereum.
The scope of the problem
To put the "75% of transactions are bots" problem in perspective, it's worth noting that the bots are not evenly distributed across all dapps. Some dapps have mostly organic traffic, while others are almost entirely driven by bots.
Also, the bot traffic clusters around the most seemingly popular dapps. That's why they seem to be the most popular.
The problem is similar to that of wash trading on crypto exchanges, which is also overrunning the crypto ecosystem, and which is likewise believed to be most prevalent on exchanges where there's no actual cost in the form of transaction fees for operating bots.
Also like wash trading, the bot activity isn't always being perpetrated by the dapp or exchange owner.
The different types of bot
In another similarity to exchange wash trading, different levels of sophistication have been observed among bots. Organic human usage is sporadic, inconsistent and generally organic-looking, while the most basic and obvious bots just wash a constant amount of transactions through a system.
These bots are there for sheer volume. They either don't care about being easily spotted or are just super simple for some other reason.
The pattern is the same for both dapp and the most basic exchange wash trading. The above picture shows those super consistent trade volumes on an exchange called CHAOEX and was presented by Bitwise as an example of wash trading.
Then on the slightly more complex side of things, you have the bots that are still pretty basic but take at least some steps to avoid detection.
AnChain gives this example of an EOS account that made transactions almost every four hours from 2 February to 8 February.
Read it in columns from left to right. You can see how there seems to have been a bit of a set-up and adjustment period over the first couple of days before it was just left to run.
When this kind of bot is used to wash trade on exchanges, you often get hypnotic and very inorganic results. It's not as easy to spot as the sheer volume bots, but it's still clearly not human activity.
Analyst Sylvain Ribes spotted the above volumes on the OKEx cryptocurrency exchange.
It's worth noting that OKEx openly acknowledged that the exchange was having problems with users wash trading but was finding it difficult to stop the activity. According to OKEx, the wash trading on its exchange was from users deliberately inflating their regular volumes to reach more competitive fee tiers from higher regular volumes.
The same wash trading problem is present on even the cleanest-seeming exchanges, it said.
The bot arms race in the blockchain
Bots and bot detectors have been locked in an arms race that predates blockchain, but it's a whole new kind of problem on decentralised networks.
"Pseudonymous blockchain transactions make it more difficult to detect and defend against bots compared to IP-based Internet transactions or KYC (Know Your Customer) accounts that are governed by a centralized authority, like ICANN, SEC, etc," AnChain explains. "Put more simply, the decentralized nature of the blockchain industry creates a much more arbitrary, and nuanced, operating environment that leaves the door open to bots going undetected... bot addresses can be replaced anytime, and often grow in rapid fashion making it difficult for manual blacklist input or flagging."
OKEx cited the exact same problem in its own fight against bot trading.
The solution, AnChain suggests, is to fight fire with fire and to use AI-based bot detection techniques – one of which it just coincidentally happens to be selling – to start detecting bot activity in a more advanced way.
Beyond that, exchanges can start levying reputation scores on users "similar to FICO credit scores" to penalise the suspected bots, it says.
Similarly, bot issues could also be a major problem in the future of blockchain gaming. What's a developer to do if they want to ensure that their competitive online game is only for humans to compete against other humans? Off the blockchain, the main solution is just to painstakingly keep banning robots one by one. On the blockchain, it's not so easy.
What a world
We've filled the online casinos with robots, who are mostly playing against the house-bot. It's not much fun for humans, but hopefully the robots are having a good time. Meanwhile, the cryptocurrency markets are increasingly occupied by legitimate trading bots. Consequently, they're increasingly being programmed to outwit other bots rather than humans.
Elsewhere, hackers have started programming parasitic cryptojacking malware to kill any competing malware that it finds when it infects a new host.
The digital world is becoming a place for us to pit robots against each other in gladiatorial contests, where they compete to earn profits for their owners.
And then you have one-offs, such as Fomo3D. It was a popular blockchain Ponzi game, which quickly turned into a place for people to create bots and pit them against each other simply because no human was physically able to keep up. But in an awesome twist, it was eventually won by a human who used the power of creative thinking to outwit the bots in a way that wouldn't have worked against humans.
No matter how you slice it, that's completely and utterly cool.
The robot uprising is being deliberately self-inflicted, and that's also pretty cool in its own way.
Disclosure: The author holds BNB and ZIL at the time of writing.