AnChain is a new analytics startup, and it’s on a mission: to uncover dApp bots wherever they hide. During Q1 of 2019, the firm surveyed TRON’s top ten gambling dApps and found a large number of bots. Roughly 31% of surveyed accounts and 19% of transactions were bot-driven, accounting for a whopping $270 million of dApp volume.
For TRON’s critics, this is an enticing follow-up to reports about TRON’s bot-driven Twitter traffic. However, AnChain isn’t pointing fingers at any single blockchain. It previously found that on EOS, bots accounted for 51% of surveyed accounts and 75% of transactions. The firm plans to examine Ethereum as well, according to a representative.
AnChain’s goal is to promote better data and bring transparency on all blockchains. Though it makes recommendations for all sectors of the blockchain industry, it’s particularly focused on rating sites. “All DApp rating sites ought to leverage sophisticated bot detection engines,” AnChain says—and it’s in a good position to get this done.
Finding Elusive TRON Bots
AnChain’s report explains that it is quite difficult to detect blockchain bots. For one thing, blockchain addresses can be rapidly generated, meaning that static blacklists of bots aren’t too useful. Plus, bots are hard to distinguish from humans if they behave in complex ways, meaning that rule-based bot detection isn’t always effective.
For instance, single bots often behave according to simple, regular, and obvious patterns. However, AnChain has observed that TRON allows multiple accounts to be created quite inexpensively. As a result, many TRON bots are “group bots,” and their coordinated behavior patterns only become apparent when accounts are examined together.
AnChain used machine learning techniques to sort through thousands of accounts and find these group bots. This has proven to be quite effective: AnChain was able to identify unusual activity that it initially missed, raising the detected number of bot accounts from 18% to 31%—a substantial increase, and one that is not easy to achieve.
It’s not all bad news: TRON’s top three gambling apps were largely human-used. On the flip side, this means that bots were concentrated among lower-ranking dApps: over 70% of the discovered bot accounts were found in the 4th, 7th, 9th, and 10th largest dApps.
Ultimately, the severity of TRON’s bot problem depends on how you slice it.
Good Stats Are Hard To Find
AnChain is an analytics firm—despite its advice, live data sites aren’t yet diving into bot statistics. State of the DApps, DAppReview, and Dapp.com do not appear to have ever commented on the presence of bots. DAppRadar, meanwhile, claims that it excludes bot data from its stats, but it’s not clear how its data has been adjusted.
DAppTotal does a better job: it removes bot data in a clear way. It offers a toggle that hides and shows bot data, and it also shows a column that displays how much dApp activity is “real.” DAppTotal’s bot-detecting systems may or may not be as advanced as AnChain’s systems, but it is making an effort while few other sites are doing the same.
In fairness, other sites are more concerned with other statistics. DApp.com, for example, has noted that humans can falsely drive up usage statistics: many daily “active” users actually log into a dApp without using it. DAppRadar, likewise, has noted that airdrops can attract inactive users.
Basically, bots aren’t the only way to inflate statistics.
If You Can’t Beat ‘Em, Join ‘Em?
As AnChain observes, trading bots were a precursor to what is currently happening in the dApp world. However, AnChain may be picking the wrong side of the fight: although many complain that trading bots are causing widespread market manipulation, bots are also useful tools for traders. Often, human beings just can’t compete with bots.
AnChain does acknowledge that “good bots” exist, but only for QA testing and gameplay purposes. It doesn’t seem to endorse the general mainstreaming of bots, a trend that Dropil, Hummingbird, and other bot services are capitalizing on. It’s possible that similar services will be targeted at dApp users, though none seem to exist right now.
Unfortunately, bots can be used for evil as well: they can create false data, mislead customers, and prop up token values in an unsustainable way. It’s particularly bad if dApp creators or investors use bots to raise a dApp’s ranking.
Thanks to AnChain, the rest of us might not need to surrender to a bot takeover—at least, not yet.