Curiosity Gaps
For the last five years or so I’ve been posting a sort of cooking and recipe diary. It started during the most restrictive Covid lockdown in 2020 when I thought it would be fun to share what I was cooking every week with other people who were trapped at home. I didn’t have much else to do.
When I could leave the house again, I switched to a monthly format, and started uploading my recipe notes too. For years I’d been keeping notes in paper notebooks, and, being an arch nerd, I realized I could do the same thing with a Git repository. There had been so many times I’d been visiting someone and wanted my recipe for scones, or someone had asked me for a recipe, and wanted me to send a copy. That’s a lot easier to do when you’ve got your notes in an electronic format.
My ambitions were (and still are) modest. It’s a way for me to organize and reflect, and maybe so that it’s easier to share ideas with friends and family.
Then in the last year or two, something flipped. Maybe my content got better. Maybe the Google search algorithm changed. In any case, I started getting much more traffic, mostly from major search platforms like Google. I’ve gone from a handful of visitors a month — presumably my friends and family coming for their monthly helping — to getting hundreds of unique visitors a month.
It’s a weird feeling to know that my random notes about making croissants have been viewed by thousands of people.
And this newfound notoriety, however minor, has given me an interesting insight into the machinations of running an online publication.
Visiting other food and recipe websites can often be an exercise in frustration. You have to endlessly scroll to find the actual recipe, and even then, it’s in a confusing and hard-to-read format. They push so hard for you to click around or sign up for their newsletter, or create more page area to drop ads, you can barely find the content you want.
Now that I’m on the other side, I’ve begun to understand why. The people who visit my site are almost entirely what I’m going to call “hit and run” visitors. They’ve clearly done a Google search for “polenta cake,” landed on my page for that recipe, scanned it for a minute or two to grab the details, and left.
I’ve intentionally made no effort to get people to stick around, and it shows. Of the thousands of people who’ve arrived on one of my recipes in the last couple of years, exactly one person has taken the time to subscribe to email updates.
Fortunately, this writing project isn’t my livelihood. I’ve neither made nor make an effort to make money from the project. (Granted, if you’d like to pay me $1,000 to feature your brand in a tasteful sponsorship deal, I won’t say no.)
For publications trying to make it as a business, I can see how this pushes them into reader-hostile anti-patterns.
I’m browsing other people’s food writing all the time. That can be mainstream sources like Felicity Cloake’s “How to Cook the Perfect…” column in The Guardian. It’s as often something like the relatively obscure Je pense donc je cuis. When I find a food writer that I like, I’ll go poking around their site, check it regularly, and stay engaged.
If most visitors are inclined to visit once and never return, and your business depends on their eyeballs or subscribing to remain viable, you have to get aggressive. This has to be why sites adopt tactics like the maddening and never-ending solicitations to subscribe or visit a related page that I find on otherwise great sites like Just One Cookbook.
What I’m less certain about is what to read into this. There's a part of me that feels like people need to be more independent and curious. Alec, the presenter of Technology Connections, shared a video in roughly this direction a few weeks ago. Take control of your own feed. Maybe try to explore beyond the obvious sources like Bon Apétit, Preppy Kitchen, and NYT Cooking.
At the same time, I feel like the real source of the “problem” may have deeper roots.
There’s so much content online, and the tools for sifting through it aren’t optimized for discovery. For all the hype, LLM-powered tools like ChatGPT aren’t really that much better than second-generation search engines like Google in terms of helping you find novel results. Google won the market by delivering more relevant results than Yahoo! or Bing. But, by dint of optimizing for and ranking by popularity, necessarily surfaces more generally acceptable (so popular) and thus somewhat bland results.
Large language models work a bit differently — much better in some ways, much worse in others — but ultimately have the same kind of “averaging out” effect. If you ask ChatGPT for a chocolate chip cookie recipe, you’re going to get an “average” recipe because that’s the probability-maximizing sequence of words in the complicated distribution the model has learned though sampling tons of writing.
This is something beyond any reasonable person’s control. It would be ridiculous to expect people to write their own search engines, and I have no idea how else — other than maybe word of mouth — that you’d find out about a random, obscure food blog. Should we really be so surprised that people all wind up going to the same handful of places?
Having worked on recommender systems, the novelty-discovery angle is something that even the best systems struggle with.
Likewise, I wonder if we may have the wrong expectations to begin with. It seems possible to me that my “hit and run” visitors are still displaying more curiosity and getting more novelty than they were in the past.
In the pre-Internet age, people relied on family recipes or bought one “universal” cookbook like The Joy of Cooking, and worked out of that. Only about 3.5 million copies of Julia Child’s Mastering the Art of French Cooking have been sold since it was published in the 1960s. That’s a lot of copies to be sure. It’s one of the best-selling cookbooks of all time. It’s also only about 1% of the US population. Even assuming you only need one copy per household rather than per person, that’s still not a ton of penetration. And, if we’re honest, I highly doubt most people who even have a copy have ever actually cooked from it.
From that perspective, the idea that thousands of people have read my relatively obscure notes is somewhat miraculous. Even if very few people get beyond the one recipe that bubbled up to the top few hits of their Google search.
I’m still going to log recipes on my site and keep notes every month, whether or not the project ever makes me a dime. My inadvertent (very limited) success has also been an interesting learning experience. It’s perhaps a good example that, for all the strong opinions and prognostications floating around, it’s hard to write the definitive story of technology’s impact in the moment.