Although CC is doing as well as it ever has, setting a new record last year with over 12 million pageviews, our current success does feel somewhat fragile. I’m not paranoid, but there are some serious threats out there, and not just to CC. The web is evolving and not in a good way, and it may soon get much worse.
The two image examples above are representative of the two main threats: garbage sites that optimize their headlines so as to appear catchy (“Click bait”) on Google Discover and drown out legitimate sites like CC and Google’s new AI Mode, which threatens to destroy the vital search links that millions of web sites depend on for traffic.
CC’s viewership comes from three primary sources:
- Direct: folks who come directly to the site, perhaps daily, or every few days or so. Direct traffic is of course the best, inasmuch as it does not depend on the other two. The reality though is that direct traffic has been tapering off for years, since about 2015, as a result of social media, the fragmentation of the web, and the use of phones as the primary viewing device.
- Search: viewers that click on a search result (most of Google) that offers CC in the search results. Due to the high ranking of CC in general and our reputable content, we generally do well with Google searches, and it used to comprise almost half of our traffic.
- Google Discover (and similar apps): Discover is the continuous feed of posts that Google curates for every individual user based on their interests. Discover has been a huge boon for us, as it can multiply the views any given post normally gets from 2x to over 50x. Our Vintage Snapshots as well as many of our other posts are regularly picked up by Discover and the resulting increase in revenue has allowed us to hire both Rich and Aaron.
Let’s deal with the Google Discover issue first, as that’s the image I placed at the top. This website (Tork) and a number of others like it spew out an endless stream of posts that are all-too often misleading, incomplete or just factually incorrect like this one. The title (“The American Compact Car That Outsold The Mustang”) is utterly false, and the text completely mixes up all Ramblers and the Rambler American, repeatedly. In that third paragraph, the author states that “AMC sold nearly 420,000 Ramblers that year” (1960); true, but that’s all Ramblers; the Rambler American sold just some 100,000 units. And that 420k for all Ramblers is still well below the Mustang’s 680k in 1965 and 608k in 1966 and 472k in 1967.
But, wait…further down he clearly points out that the Mustang sold in numbers well in excess of the Ramblers. So much for that headline. But no, The Rambler American was very much not “the compact car king in the US”.
It appears that all or most of these sites are based outside of the US, created simply to suck as much revenue from Discover as possible, as undoubtedly no one would go directly to these sites and they do not do well in Google searches, since Google has increasingly become much more demanding about what sites they refer in a search. Google has changed their standards several times in the past two years, causing havoc to untold numbers of sites that totally or largely depend on searches. These “HUC” (Helpful, Useful Content) standards are intended to improve results for sites that have genuinely helpful and useful content and not just SEO (Search Engine Optimized) content is is all-too often mediocre at best. This is a good thing, and has benefited CC. Yet Discover does not seem to have the same or any HUC standards that it applies to searches.
But that may all be water under the bridge, as Google has unveiled Google AI Mode, which uses AI to answer all queries instead of actually showing websites that offer it. For instance, if you search “1966 VW 1300 Beetle”, CC’s post on that subject is at the top of the page, as it has been for quite a few years. And that is the case for a large number of our posts.
But in AI Mode, a whole article about the subject is generated (only partially shown here). They do show some websites as either their source material or additional sources. But look at the top one on the right there: It’s that utterly garbage AI-based website we featured here a while back because their purple AI-generated images were often so off and bizarre. This is really bad, not only because it (almost) completely cuts out traffic to sites from searches (who’s going to bother clicking on those sites on the right when Google AI gives a complete article?) but the top link they offer up there is 100% AI-generated and not trustworthy.
BTW, Google AI does put this disclaimer on the bottom of each result, but that’s not exactly helpful. There’s no feedback mechanism or other way to tell them that it’s wrong. And if they’re using AI-generated sites as an actual source, well then mistakes are inevitable. FWIW, I did not find any mistakes in this Google AI article on the ’66 VW, and undoubtedly AI is getting better all the time. Who needs websites? Yet it was Google that created this whole ecosystem in the first place and made them so huge.
All of this is rather disheartening. We’re drowning in a world of misinformation, fake news, deliberate deception, polarization, silos of news and information, and the consolidation of the tech giants over increasing amounts of what folks are offered and consume. One might think our little corner of automotive history might somehow escape that. Well, we’re determined to do so as long as possible. AI responses will never equal the personal insights, experience, pathos, judgment, humor, irony, and other human qualities that are present in genuine human-created content such as it is on CC. That’s why from the get-go CC was always more than just about regurgitating commonly-available facts.
With your support we will carry on for as long as possible. But there’s a cold and chilly headwind blowing.
Just another example of the AI hype. My son has a masters degree in computer science with specialization in data science. He says AI is a sophisticated bullsh*t artist. He demonstrated to me how he could get Chatgpt twisted around to where it admitted it was wrong.
Caveat emptor.
Thank you for this information. I have been using AI for quite a while.And was happy with the consolidated results.But now I can see how it can be very erroneous. I especially use AI for automotive type.Searches thinking that that’s an area that is quantitative and unlikely to be incorrect. I was wrong and will be more careful and will attempt to go direct to websites like yours in the future.
MANY years ago, as a HS teacher, the fascination of even preteens with video gaming exceeded their interest in classroom learning. At that time, I became leary of the future influence of what is now AI. I couldn’t believe that students were allowed to use personal computers for math classes. Asking the head of the Math department about it, he replied that students no longer needed to waste time LEARNING information, just knowing how to find it. Was this the beginning of the end or the end of the beginning? I certainly don’t know!
I actually follow CC mostly through it’s RSS feed on Feedly (which shows about 580 followers of the feed at the moment)
I don’t remember how I found it originally, but, have enjoyed reading your articles for a number of years.
The other automotive site in my Feedly feed is Autopian.
I really enjoy reading articles by obviously real people. The articles here and there feel like information from a friend who happens to know A LOT about cars.
I’ve recently realized I’m not as much of a car expert as I once thought I was. I have a general knowledge of wheeled thingys going back to my childhood in the ’60s, through my years of wrenching on my own $200 cars because that’s all I could afford.
The articles and comments here show me there were a lot of details I missed.
The historical angle of many of CC’s articles is very interesting to me, especially when I see an article about one of my old cars, or cars my Dad or Uncles had. The information is presented from a perspective that will be difficult for AI to duplicate – the old “Been There. Done That” which helps to solidify a reputation for expertise that people desire.
AI might get to a point where it can provide factual articles and stats about cars, but it will never provide the rich commentary and stories of the cars we have owned individually. My real interest in this site has always been the COALs, of which I’ve contributed a few. Those very human stories of cars we’ve owned or known are never going to be the domain of AI. I know CC’s original mission statement was to document the forgotten, ordinary cars of yesteryear, but I think that it evolved into doing much more than that, into telling the story of some of those cars and how they affected individual people.
Garbage in, garbage out. This old data processing adage still applies today.
“Love it! I’m pretty blown away by these (AI renderings). And the others we’ve seen here. Who could have anticipated this then years ago?” — Paul
You once embraced the power of AI.
Until, you saw that it could undermine your business model?
Link:
https://www.curbsideclassic.com/blog/what-if/what-if-1975-studebaker-lark/
I think sensible people who are able to use their brains and are interested in automotive texts will have CC as one of the most important bookmarks at the top of their browsers.
From the noughties onwards, there was a joke in my circles: “Try >Gockel<*, but remember, it's a SEARCH-engine, not a FIND-engine."
Now this company (like many others) have built an LLM function into their system, which has been fed with everything they could get their hands on – disregarding all copyright laws.
The result: we can now watch humanity go stupid live.
Because there is nothing intelligent about it. This are trained neural networks – and are only referred to as ‘artificial intelligence’ (AI) by marketing people, journalists, politicians and other digital illiterates.
Whether it could be an enrichment to have people who use AI out of ignorance on CC I cannot judge.
But based on my observations in my personal environment so far, I would say: no, probably not.
* This is pronounced similarly to "Google" in German, but means rooster.
I’ve noticed increasing trouble getting to legitimate sites and/or articles for gardening, too. Many of the articles that come up early in search results are clearly AI-generated waffle, that may or may not be approximately correct. For now, I can get through to legitimate websites with a little persistence, but as Paul says, how many people will bother to check out links of any kind if all the “facts” are neatly summarized for them at the top of the page?
As AI continually trains on itself, more and more errors will be perpetuated and magnified. There are even predictions that in a few years it will be impossible to tell what is real or unreal. Younger people who grow up going to AI for answers to everything will not even have developed critical reasoning skills, much less research skills. It’s all quite depressing.
AI isn’t inherently the problem here, it’s how Google positions it as the lead result. Similar would occur if they positioned Wikipedia as the default lead result. It gives people a “top” result which many will read and not go any further. If Google’s AI was repackaged as “Googlepedia” it’d be even less transparent.
Wiki, as we well know, is far from perfect with plenty of outright falsehoods in it. Yet it serves to give information that suffices and is “correct enough” what, 98% of the time? And if one is interested in a deeper dive (or to verify it all or for something more obscure) then one seeks out specialty sites and sources, sometimes a number of them.
Google’s AI results annoy me to no end as it is so often wrong about everything, usually blatantly so but that assumes I know enough about the subject already to realize that. I understand that and generally ignore it and scroll down, but obviously there are plenty of people that just eat it all up as with many other things out there to everyone’s detriment, AI itself isn’t causing human idiocy but it is furthering it. One by one people may realize it is not perfect and question it more and more. But if the creators won’t police it (or have a way for users to police and factcheck it), then the whole system/society is on its way down as we’ve been seeing happening even before AI with blatant falsehoods being propagated on purpose and not effectively stopped (politics and “news”). At the end of the day Google did create their ecosystem, it’s their playground, we just use the swingset. Until they want to hog it all the time.
CC is lucky in that we got here before AI, so we created much of the body of work (or A body of work anyway), and that is there as a reference for those sentient enough to realize it is needed, however AI will get better and better, creating less of a need or desire for it. But perhaps it won’t get better, AI can just as easily use sites such as the junk sites for reference, which may just as easily be AI generated themselves (from a different engine) and then keep consuming its own junk. As you stated, much of the appeal of the site is in things other than just raw facts, the challenge is in how to get that in front of people so they realize this site as a resource. The “feed” is algorithm (which IS AI) based, obviously, and it’s providing stuff we offer to people that aren’t overtly asking for it, it’s not a big step before it generates and organizes that content itself as well. Then, as far as this website and others are concerned, we become the magazine industry of the last 30 years.
Another car website did a test where they asked an AI engine to write a post in the voice of several of their authors, they (those authors) then picked the results apart as far as things and phrases they would never use, but you know what? On the face of it, I didn’t immediately suss out that it wasn’t their writing and i doubt many others did either. I’m a little afraid to check if C3PO and his buddies could write some of the things I write late at night while giggling to myself but I’m sure they could get uncomfortably close…
The challenge is in how to repackage it as a force for good instead of evil (and mis/disinformation IS evil, rotting society from the core). However I don’t see the collective will out there to do that, so I unfortunately do not have the answer.