Does Reddit have an astroturfing problem? What the data suggests
Concerns over astroturfing on Reddit are widespread, with many users citing paid comment services, political influence campaigns, and brand shilling as evidence that large portions of the platform’s apparent “conversation” may be manufactured. Commenters argue that Reddit’s business incentives, weak enforcement, AI-generated content, and new features like hidden comment histories make manipulation easier and detection harder, undermining its reliability for product advice and political sentiment. Some advocate returning to smaller forums or niche communities, seeing mainstream social platforms as increasingly overrun by bots, spam, and coordinated propaganda.
Perceived scale of astroturfing on Reddit
- Many commenters assert Reddit has substantial astroturfing across politics, brands, crypto, gambling, and product recommendations.
- Several describe obvious coordinated campaigns: niche products pushed with identical phrasing, political brigading in local and national subs, and necro-bumped recommendation threads.
- Some say Reddit is now unusable for product advice; others estimate a large share of discussions as “fake or fabricated.”
- A few admit to deliberately nudging political discourse themselves and find it “terrifyingly” effective.
Business incentives and platform behavior
- Multiple comments draw parallels to Amazon reviews, Google/Facebook scam ads: platforms need engagement/content and are disincentivized to crack down hard.
- Suing shill companies is seen as impractical: low provable damages, operators can be overseas, and new ones appear.
- Some speculate Reddit benefits from shill activity (engagement, proof ads “work”) and may even be lenient or financially aligned, though this is unproven in-thread.
- Allowing hidden comment histories is widely viewed as making shilling easier.
Detection signals and their limits
- Common heuristics: young or “Word-Word-####” accounts, thin histories, repeated promotion of one brand, and store/affiliate links.
- Others argue these signals are now weak: bots can age accounts, post “lost” content in unrelated subs, and mix real-looking comments with occasional ads.
- Hidden or wiped histories and the rise of convincing LLM-generated posts make the “Turing test” feel obsolete.
- Account age is seen as one of the few hard-to-scale signals, but it excludes many legitimate newer users.
Impact on trust and behavior
- Many no longer trust Reddit for product reviews; some only use it for general “what to look for” guidance or niche technical questions.
- Several report abandoning local or interest subs due to bots, ragebait, or heavy-handed/agenda-driven moderation.
- Some now assume every commenter might be a bot or shill and read everything critically.
Comparisons and alternatives
- Users mention curated review sites (GearLab, OutdoorGearLab, Consumer Reports, Rtings, Wirecutter, Project Farm) as relatively better, but debate sustainability and susceptibility to affiliate bias.
- There’s nostalgia for traditional forums and smaller, gatekept communities as less polluted by bots and astroturfing.
Critiques of the linked analysis
- Many find the article’s AI-generated prose hard to read and its methodology narrow or statistically weak.
- Some think finding bots for a single smaller brand suggests a broader problem; others say the study “fails to find” what it implies and is poorly designed.
- The prominent use of LLM drafting is itself criticized as emblematic of a “dead internet” full of synthetic content.