Is the Internet Actually Dead?
The Dead Internet Theory started as a fringe forum post. The bot-traffic data behind crypto's shilling economy makes it a lot less fringe than it used to sound.
Is the internet dead? And more specifically, is crypto’s corner of it being run by bots pushing coins up and dragging others down?
What The Bot Traffic Data Actually Shows
The Dead Internet Theory started in 2021 as a fringe forum post arguing the real, human made internet had quietly died years earlier, replaced by an ocean of automated content. It got mocked for years. Then the traffic numbers started catching up to it.
Imperva’s 2025 Bad Bot Report found automated traffic at 51 percent of web traffic in 2024, the first time in a decade that bots outnumbered humans, and its 2026 report puts 2025 at 53 percent, with about 40 percent of all traffic classified as bad bots specifically, not search crawlers or legitimate monitoring tools. On June 3, 2026, Cloudflare CEO Matthew Prince said Cloudflare Radar showed 57.5 percent of HTTP requests worldwide coming from automated systems, while admitting the classification is still a bit messy. On X specifically, estimates of automated accounts vary enormously, from X’s own claim of under 5 percent of monetizable users to academic estimates of up to 15 percent of active accounts in 2017, and nobody has reliably measured crypto heavy reply threads.
It’s worth being fair to the skeptics here too, because the caveat matters. A web request isn’t a person. A single human loading one webpage triggers dozens of automated requests for images, scripts, and fonts, so raw request percentages overstate how much of the actual conversation you see is fake. Several researchers have also pointed out that vendors measure bot traffic differently, so the numbers are hard to compare across sources, and that the share has moved up and down over the years. Even so, Imperva’s own figures swing from 61.5 percent of traffic in its 2013 report down to 37.2 percent in 2019, and then back up to 47 percent in its 2023 report and 53 percent in its 2026 report. So the strong version of Dead Internet Theory, that literally nothing you read online is human anymore, isn’t proven. The weaker, more defensible version, that a very large and rapidly growing share of what looks like organic activity is automated, is well supported and getting harder to dismiss every year.
Crypto’s Bot Problem Is Not Hypothetical, It’s Studied
This is where it stops being a vague internet wide vibe and becomes a specific, documented pattern inside crypto.
In a 2021 study, researchers at USC’s Information Sciences Institute tracked cryptocurrency pump and dump operations directly on Twitter and found something stark. About 80 percent of the social media conversation around a pumped coin was initiated by bots, not real participants reacting organically. The fake accounts were often laughably easy to spot once you knew what to look for; many shared the exact same dog photo as a profile picture and were all created within the same short window of time. A separate academic study of pump and dump mechanics documented the actual choreography. Once a pump starts, coordinated members post about the coin, retweet the project’s own account, and spread claims in Telegram and Reddit communities without disclosing that a coordinated scheme is underway, while bot accounts with large follower counts amplify the volume until exchanges and tracking sites start listing the coin as a top gainer on their own, purely because the manufactured activity looks like real demand.
The negative side of this exists too, and it’s just as documented, if slightly less discussed. Longstanding allegations, going back years, accuse major stablecoin issuer Tether and its affiliated exchange Bitfinex of running or benefiting from bot networks, with individual users publicly noting the same suspicious accounts appearing to promote Tether affiliated projects while attacking critics. Separately, in 2018 independent data analyst Geoff Golberg mapped follower networks around prominent XRP accounts and found clusters of apparently fake profiles, though that analysis was not peer reviewed and says nothing about the many real XRP holders. Some of this is documented and some of it remains allegation, but the tactic itself is well known. It’s the kind of astroturfing that researchers studying crypto specifically have named and mapped as a common, recurring marketing tactic across the industry, used both to inflate support for a favored project and to manufacture the appearance of grassroots hostility toward a competitor.
The Short Selling Incentive, The Actually New Part
Once you can short a cryptocurrency as easily as a stock, the financial incentive to run a coordinated negative campaign against it stops being abstract.
The pattern already has a name in traditional securities law: short and distort. An operator opens a short position, then floods social media and forums with negative claims, often framed as legitimate research but built on selective or false information, timed to trigger a panic sell off before quietly closing the short for profit once the price has cratered. It’s explicitly illegal under existing securities fraud statutes when it happens with stocks, and regulators have pursued real cases over it.
Academic researchers have now built a specific model for detecting this exact pattern inside crypto markets, publishing a paper that identifies short and distort manipulation in cryptocurrency trading using measurable signals: price reversals, abnormal trading volume, and a widening gap between buy and sell prices, distinct from the signature of a traditional pump and dump. The fact that researchers built a detection model for it means the pattern has already been observed happening, not just theorized.
And this isn’t a purely academic concern sitting in a journal somewhere. In 2025, Michael Saylor, executive chairman of Strategy, one of the largest corporate holders of Bitcoin, publicly claimed the company’s harshest online critics weren’t real analysts or retail investors at all, but a coordinated bot network funded by short sellers specifically to amplify negative sentiment and drive the stock down. Veteran short seller Jim Chanos challenged Saylor to produce evidence, calling it a serious allegation stated as fact, and noted that skeptics like himself have been transparent about their reasoning. Whether Saylor’s specific accusation is accurate is a genuinely contested, unresolved dispute, and it should be treated as one rather than settled fact in either direction. But the framework he’s describing, someone with a financial incentive from a falling price allegedly deploying bots to help it fall, is exactly the mechanism securities law already has a name for, and exactly the mechanism academic researchers have already built detection tools to catch inside crypto specifically.
Putting It Together
So, is the internet dead? Not entirely, and the people who say it’s just evolution rather than an apocalypse have a fair point. But the specific idea at the center of all this, that crypto in particular has a substantial, documented bot problem, both inflating coins its operators want to succeed and, increasingly, degrading coins that someone somewhere has a financial incentive to see fail, isn’t speculation. It’s published research, with real academic detection models built specifically because the pattern was already showing up in the data.
The genuinely new wrinkle is the short side. Pump and dump bot activity has been studied since at least 2019. The idea that the same bot infrastructure gets pointed in reverse, to profit from a coin’s decline rather than its rise, is the part that’s caught up to crypto only recently, mostly because shorting crypto easily is itself a relatively recent development. Give that incentive a few more years to mature, and the negative campaigns are the ones worth watching closest, precisely because nobody’s watching them as closely yet as they’re watching the pumps.