Because it's the most secure app I can actually get people to install.

If you have friends or family privacy conscious enough then sure, use something different, or shoot, even self host. But for everyone else, even getting them to use Signal is a difficult feat.

Yes it's not private who you chat with, but the content of the messages are secured and the keys are not stored in a way that's accessible to the central server like most "e2e encrypted" messaging systems. So yes, it's not perfect for journalists and similar people who need to hide the who, but for those of us who just don't want the content of our messages scraped for advertising, anti-minority attacks, or any other reason, it's a convenient, reliable, and well designed system. High availability and ease of migrating my contacts is what's most critical after security for me or my contacts would abandon it. Most can't afford the time, effort, and server costs for high availability of a self-hosted application. And what if that friend who self hosts loses their job or just decides to stop hosting it?

And truly private applications that don't have user IDs make it really difficult to find people when you switch devices so never have really caught on. It's hard to get nontechnical people to put in the effort to prevent having to start from scratch reconnecting with contacts offline to get their online identities every time something breaks. Not to mention if they get popular they will get flooded with tons of spam without a way to limit users to a single identity. It's just very inconvenient and thus for most people not a good tradeoff.

That's the neat part: you don't.

deltachat!

For most people, signal is the first step.

it was hard enough persuading people to talk to me on signal rather than FB messenger/whatsapp etc

Because it's tested. Seriously. Maybe it's compromised in some way, but we haven't seen any evidence of that yet, rather we have seen cases where it was very good at protecting people. You may be suspicious, but it can be a mistake to treat suspicion as fact. The problem with a lot of the alternatives, is that they haven't been so broadly 'tested'. We don't have enough data to really trust them IMO.