Perplexity, My Server Logs, and Who Gets to Be Mad

Perplexity, My Server Logs, and Who Gets to Be Mad

Tech News ai blogging journalism perplexity web-scraping

So I've had three separate people email me this week asking if I'd seen the Wired piece on Perplexity, which, first of all, hi, yes, I read things other than my own blog occasionally. But also, yeah. I read it. "Perplexity Is a Bullshit Machine" is a genuinely great headline and the reporting underneath it holds up: Wired caught the AI search company's chatbot fabricating details in a story about a police drone program, complete with quotes that don't check out, while Perplexity's traffic to Wired's own site apparently came from an IP range that had no business ignoring their robots.txt file the way it did. Forbes ran into basically the same thing a couple weeks earlier with a story about Eric Schmidt's drone startup, where Perplexity's "Pages" feature spat out something that was a little too close to Forbes' own reporting for anyone's comfort.

I want to be annoyed on principle here and mostly I am, but I'll admit my annoyance is personal, not abstract. I've been doing this blog since November 2011. Thirteen years, on and off, of writing stuff, hitting publish, and watching my server logs fill up with bots. Googlebot I don't mind, that's the deal, that's always been the deal, you let Google index you and in exchange people find your posts through search. Fine. What's happening now is different. These AI crawlers aren't sending anyone back to me. They're eating the post, chewing it up, and handing some paraphrased sludge to a user who never sees my name, never clicks through, never knows techpad exists. I checked my own logs Friday night out of pure spite and found a user-agent I didn't recognize hitting old posts from 2015 at a rate that made no sense for a human. Could've been anybody. Probably wasn't Perplexity specifically. Didn't matter, I still felt a little violated about it, which is a weird thing to feel about a WordPress install nobody reads but me and my mom.

Here's where I won't pretend to have a clean opinion though: I also think a lot of the publisher outrage is coming from outlets that spent fifteen years scraping and aggregating everyone smaller than them without a second thought. Forbes! Forbes is upset about content being lifted without credit! The same Forbes whose contributor network has been a running joke in SEO circles since roughly 2013! I don't feel bad for them specifically. I feel bad for the reporters who actually did the drone story, whose byline and paycheck depend on people reading their site directly, and who are now competing with a chatbot that ate their work and handed it out for free with the seams less visible. Those are two different things and it annoys me when people flatten them into one outrage.

Aravind Srinivas, Perplexity's CEO, has mostly responded to all this by saying the company is exploring ad-revenue-sharing deals with publishers, which, sure, maybe, ask me again in a year. I'm not holding my breath. Robots.txt is a polite request, not a lock on a door, and it's been treated as optional by somebody every few years since I started paying attention to this stuff, it was ad-tech scrapers in 2012, content-farm bots in 2017, now it's an AI company with a $3 billion valuation and a chat interface. The tool changes. The instinct to just take the thing that isn't nailed down does not.

Unrelated tangent because I don't believe in tidy blog posts anymore: I've also been half-following the Boeing Starliner situation, where Butch Wilmore and Suni Williams are still stuck up at the ISS well past when they were originally supposed to come home, because NASA keeps finding new helium leaks and thruster issues and keeps pushing the return date back another few days. Every time I check the news there's a new "targeted no earlier than" date. At some point that stops being a schedule and starts being a vibe. I don't have a joke here, I just think about those two a lot lately, sitting up there while people on the ground argue about valve seals.

Anyway. I'm going to go look at my logs again. Somebody's definitely reading this post about people reading my posts without reading my posts, and honestly, good for them, I guess.