Yeah, I feel like the generative AI Google has added (for those who opt-in) has been a pretty useful enhancement. It will read through the results and try to summarize it back to me in a useful way, the same kind of summary I’d typically try to find by including Reddit in the search term.

However…it’s pretty common for the AI result to get some of it’s information from Reddit postings. Like this morning I asked “on myfitnesspal is the fitbit calorie adjustment accurate?” and it had some of the generic chat answers about how calorie estimation is done in general in exercise science, but the most useful bit was the last sentence claiming that “MyFitnessPal overestimates calorie needs in 95% of cases because it overestimates calories burned from activity”.

That’s a pretty broad conclusion, and the first supporting link it provided was a detailed 8-year old reddit posting from a software developer examining the MFP calculation, comparing against measured study results on calorie burn estimates by activity, and noting a consistent overestimate by roughly 20-30% which appears to fall in line with the baseline calorie burn that users would experience even if they did not perform that exercise activity. His conclusion was that the estimate is directionally reliable if you deduct 20-30% for the double-counting of sedentary calorie burn.

That kind of in-depth examination by someone with expertise and without financial motivation is the contribution of the “hive mind” the critical mass of people serendipitously stumbling across topics/questions they have answers for and volunteering useful information. We’re going to lose a lot of that unless or until some replacement for reddit pops up. A diaspora of small niches of information may not be as useful if it is buried on bot-generated clutter in a way that search isn’t able to sift signal out of the noise.