Because that's how surveillance-as-a-business-model[1] works[2]. The industry that brought us the "EULA" just loves to try to force contracts on people, especially when it tricks the user into giving up even more data.
It starts as a useful product until the network-effects take over and raise the cost of leaving, de facto locking people in. After they are hooked, you introduce various types methods to make the product rely more on remote services instead so more data can be captured. As few people understand the difference between a product that runs locally and a service that necessarily gives your data over to a 3rd party, this usually works without anybody noticing and objecting.
If a few nerds object, they are usually countered with a few lies about why such a service is "necessary", even when it isn't. In this case, the lie is that a 3rd party service is necessary so you can share your bookma^H^H^H^H^H^Hreading list on different devices (which assumes you have multiple devices, and that you want to move bookmarks between them). In hard cases, it may require some vague, misleading, and hard-to-prove statistics ("our data says user's like $foo", "everybody uses $foo"), or simply browbeating anybody that complains ("Then submit your own patches", "stop being paranoid").
The worst part is that for many of the engineers involved, this is probably unintentional. The human mind has a very limited "working set" that is given full attention; everything else is filtered heavily with various shortcuts[3], heuristics, and assumptions, making it incredibly easy to be distracted by endless technical details.
A more plausible explanation is that that's the only way people discover new functionality. Prediction: it won't be the default eventually (after people get used to the idea that they can easily pick and choose sharing services https://activations.cdn.mozilla.net/en-US/ to show).
It starts as a useful product until the network-effects take over and raise the cost of leaving, de facto locking people in. After they are hooked, you introduce various types methods to make the product rely more on remote services instead so more data can be captured. As few people understand the difference between a product that runs locally and a service that necessarily gives your data over to a 3rd party, this usually works without anybody noticing and objecting.
If a few nerds object, they are usually countered with a few lies about why such a service is "necessary", even when it isn't. In this case, the lie is that a 3rd party service is necessary so you can share your bookma^H^H^H^H^H^Hreading list on different devices (which assumes you have multiple devices, and that you want to move bookmarks between them). In hard cases, it may require some vague, misleading, and hard-to-prove statistics ("our data says user's like $foo", "everybody uses $foo"), or simply browbeating anybody that complains ("Then submit your own patches", "stop being paranoid").
The worst part is that for many of the engineers involved, this is probably unintentional. The human mind has a very limited "working set" that is given full attention; everything else is filtered heavily with various shortcuts[3], heuristics, and assumptions, making it incredibly easy to be distracted by endless technical details.
TL;DR - watch [2] for a much better explanation
[1] aka "big data" and sometimes "analytics"
[2] https://projectbullrun.org/surveillance/2015/video-2015.html...
[3] most illusions and magic tricks are based on this - Apollo Robbin even bases his entire show on working just outside the focus-set of his audience ( https://www.youtube.com/watch?v=T0k2gja3ym4 https://www.youtube.com/watch?v=d54ydsKUNGw )