|
that "Google made their search results worse on purpose to get more ad impressions" article that's going around is based on an internal email chain from a legal filing, in which an employee mentions this potential approach for increasing engagement metrics... as an example of a bad approach and not something they'd be comfortable doing. based on my reading, there's nothing in here that says they actually did it. I can't see anything in there that supports that claim. 40 comments
certainly unsatisfying that Google is not unambiguously the bad guy in this case, but since there's no actual evidence of malfeasance it's just straight up misinfo the fact that search quality seems to have dropped in the months after this email chain doesn't concretely support a causative link or intent. the push towards AI summaries likely impacted a lot of that. do I think it was pure incompetence? probably not. do I think it was 100% premeditated malice? also probably not. my gut says they expected AI summaries to increase engagement, and then they did but not for the right reasons, and then multiple incentives coincided so they kept it. maybe it'll turn out they really did do it on purpose. wouldn't put it past them. but for now we have no solid evidence that they did, even if that narrative would fit perfectly with public perception and make for great headlines. still though, fuck Google, for so many other reasons. just probably not this specific one. oh and duh, I should've said which post I was referring to. the post: https://raru.re/@ocean/114490739719068242 the article the snippet is taken from: https://journalrecord.com/2025/02/20/is-google-making-search-worse-to-sell-more-ads/ @danderson @gsuberland the second half of the problem is that search quality is a problem becoming harder over time, so even if you maintain the same level of investment your quality goes down. (Reasons include: expectations rising, amount of content on the internet rising, SEO being more and more adversarial, giant information silos like FB/Discord, paywalls, abuse profitability rising, and probably dozens more.) @gsuberland is that the Where's Your Ed At article from April '24 (https://www.wheresyoured.at/the-men-who-killed-google/)? That one points to changes from '19 which was well before the GPT and AI summary stuff AFAIK. @az no, but that article has its own problems and also fails to present much in the way of real evidence. @gsuberland also we shouldnt forget that search quality gets worse because people over time figure out how to SEO their shitsites @gsuberland its frustrating, like with the soundcloud ToS bad faith reading thing. i had someone back up to a corner of "well you cant trust companies anyway" like guh @gsuberland Ed Zitron went into a lot of detail regarding this a bit ago if you want to read: https://www.wheresyoured.at/the-men-who-killed-google/ @computerywar right, that essentially says the same thing: the email chain is suggestive of there being a push towards lowering search quality to increase engagement metrics, and there was pushback. search quality generally dropped, and an insinuation is made that it was intentional and causative, but no actual evidence is provided to support the claim. @gsuberland I think it's quite obvious at least that SEO slop is winning against Google. Before LLM at least it's hard to come up with a reason why it would be the case without Google being complacent.
@gsuberland the google employee who did that study flatly denied this too on bluesky, btw https://bsky.app/profile/moultano.bsky.social/post/3lowrooqpw22i @gsuberland oh wait no that was about a different study. still (the judge claimed that google's numbers found that worsening search results would "only" have a 1% drop in revenue when according to him the actual conclusion was that ranking quality is really important if it can affect revenue as much as 1%) @gsuberland also that survey is abysmal. It was done by a random credit card comparison company, they don't state a methodology other than "online lol" (on 2 different dates 6 months apart, all mixed together?), and the questions are basically all leading questions.  @gsuberland Looks like @sjb found this relevant quote in a Texas @gsuberland @catsalad @sjb wasn't there a talk years ago about how bigG didn't know how their ranking system actually "behaved"? I remember this coming out a few years back too (I may be mistaken) which also pushed for SEO related discussion and doubts. https://www.seroundtable.com/google-dont-understand-rankbrain-21744.html @gsuberland @catsalad @sjb Occam's Razor. @gsuberland acute observation, as always. @ciaranmak that goes without saying. just annoying that convenient but unsupported narratives get in the way of actually holding them to account. @gsuberland The alternative is they made it worse for no reason to push an unclear AI adoption agenda. @ciaranmak if the two are related, my guess is they pushed the AI features expecting engagement growth because users like it, then they did get engagement growth but it was actually because the AI summaries suck and people keep having to dig further to find and verify information, and then they figured "well ok that didn't go exactly as planned, but since the metrics are up and we really want to sell the AI stuff..." - I can see that being hard to argue against when everyone's incentives align. @gsuberland regardless of whether the article is legit, it's hard to escape the fact that google searching, and search results HAVE gotten far worse. so it makes for a believable story. Thanks for the deep dive into the source material though, it does help to know what facts vs. rumors. @gsuberland i recommend you listen to Cory Doctorowâs podcast âUnderstood: Who Broke The Internet?â, specifically the episode âDonât Be Evil.â it basically covers everything youâre calling into question. @andrea_smandrea yes, I have read the article version of that, and it provides nothing to contradict what I wrote here. claims are made, no evidence is provided, and there are a few bits that look to be patiently wrong in hindsight. @gsuberland I swear they did and then rolled it back, but I donât want to read all of that again. My takeaway was a sharp reminder that google, like meta, are advertising companies that have various web services as well. The email chain is sad. Like the slow death of a passion for technology getting fisted to death by greed. Like bowels perforated dead. @elebertus I think you're thinking of the thing where they deranked Wikipedia as an experiment for a few weeks as an experiment? and it hit their revenue by 1%, which people took to mean "they decided they could lower quality without losing revenue", but then it turns out 1% is considered a huge loss at that scale and the actual outcome was "we are definitely not doing that". @gsuberland I think this way of thinking also tends to anthropomorphize the company too much. When large organizations do something truly bad or stupid, it often arises from bad incentive structures and everyone involved âjust doing their jobâ. I think itâs important to acknowledge that shit like this can happen without any actual bad intent or conspiracy, just out of bad organizational practices and lack of care. |
Napping.Synth.Download
the article links to the results of a survey on people's perception on the quality of Google results over time, showing a negative trend, and I think that's something we all share similar anecdotal feelings about. but then it goes on to make the direct claim that they did it on purpose, which is (as far as I can tell) a claim without any concrete basis.