I strongly agree with this, with the one caveat that right now algo doesn't provide seamless integrated support yet for a VPS provider that offers a flat bandwidth cap (like OVH or Scaleway) vs a high burstable data cap oriented offering like DigitalOcean. The flat bandwidth (generally 100 Mbps on the cheap plans) tends to come at the expense of burst/cpu/disk storage, but none of those matter in VPN vs reliability and not having to ever think about going over limits, even if you want to let family members for example use it. While for a lot of general projects I'd definitely agree with their current easy cloud choices, for this particular application, for most people, I think the likes of OVH or Scaleway or similar would be a far better fit, though I realize the major holdup is Ansible support. Of course, it can still be setup wherever, just without the same ease of use for someone only mildly technically oriented which is how it truly excels right now.
Still, I think it creams every general public offering. I agree with fictioncircle above that the "anonymity" thing is a total red herring. VPNs in this application are fundamentally about creating a hack to let individuals change their Internet access from a natural monopoly situation to a strongly competitive and customer oriented market situation via virtual end point shifting. That's "it", though it's a big deal. But "anonymity" is a far, far trickier problem, requiring not just extensive infosec but also significant opsec. At a bare minimum most people would need to use something like the Tor browser, not just for the "tor" part but for the hardening they put into the browser to make it somewhat harder to get tracked anyway regardless of IP address. I think a lot of the "anonymity" marketing claims some public VPNs make verge on not merely disingenuous but outright dangerous to the extent they can create a totally false sense of security.