[riffraff]

I would argue the problem is not the broken tags, but the business disadvantage to exposing semantic data.

Remember when microformats were all the rage, and you could get hReview or hRecipe or XFN data everywhere?

Then every host in turn realized that actually, it's _better_ if people can't scrape your site, and it's even better if they can't even see it and it's behind a login wall.

[acdha]

“better” is too strong: in many cases, structured data is not a problem (and if it is, people will scrape it anyway), but there's simply no business case for spending time on it. Most of the semweb stack had a horrible developer experience — bad documentation, tools, validators, etc. — and rarely had tangible benefit from spending time slogging through it.

The semantic data which has actually been implemented on a wide scale happened because someone could go to their boss and say “Spending time on x will mean better Google ranking” or “Facebook will use their new sharing display for our pages”, and it was orders of magnitude simpler to implement so the time and risk were far more palatable.

[zozbot234]

Well, whether it's better depends on local incentives. But it's true that in many cases these push against making machine-readable data available, thus "semantic" tech becomes mostly irrelevant. Similarly, Linked Data has been most successful as Linked Open Data, where these incentives are explicitly aligned.

[gwern]

Indeed. Why would you expose all of your data to your competitors like Google, so they can commoditize you? (Incidentally, note that the big tech companies like the search engines are some of the major proponents of microformats, like for restaurants or local businesses... As always, 'commoditize your complement': https://www.gwern.net/Complement )

[chongli]

That’s a proximal cause. The root cause is that the Internet is not free, despite appearances. If hosting and bandwidth were free, we wouldn’t need businesses to do what we want. Wikipedia wouldn’t need donations. Everything would be great.