TLDR; I no longer consider google trustworthy, Tiny Tiny RSS is a suitable, self hosted, replacement for Reader.
Until this week, Google had managed to convince me their services were trustworthy – more trustworthy than self-hosting – which is quite a feat , since I don’t tend to do well with faith in any context. Killing reader after it drained the rest of the RSS aggregator market took care of that illusion. Kicking the ad blockers out of the play store (on the same day) after Android had become the dominant species, and it no longer mattered that ad blockers are required to make the mobile web experience tolerable, and
intentionally breaking Jabber federation later in the week just underscore the point.
Last time Google made me nervous, I wrote about working out a Google Exit Plan, I now find myself needing to put at least part of it into action, on a “which parts do I do now” basis.
The first order of business is, naturally, replacing Reader. I looked at a whole bunch of options, which the internet has been suggesting in droves since, and eliminated most because I’d rather self-host than just switch my dependency to another service. Newsblur and Tiny Tiny RSS were the obvious front runners. Newsblur pretty, but the documentation for self-hosting it is such that it isn’t even clear that it is possible (and I see several sites who really don’t appreciate its attractive scraping features). Selfoss also showed some promise, but doesn’t look quite finished enough to start using today.
TTRSS won on several points – it has relatively simple hosting requirements (*nix, http server, PHP ≥ 5.3, and My or Postgre SQL, is flexible and transparent, and has a nice Android app (two actually, forked from a common code base).
My Bluehost shared hosting is usable but not ideal for ttrss – it has php 5.2, 5.3, and 5.4 binaries for cgi-fcgi, so it will run, but only 5.2 for cli, which means background updating is kludgey in an abusing-wget-and-cron sort of way. I’ve never particularly enjoyed that sort of web work, but it was straightforward enough to get going, performs rather well, and with only a few settings changes and a quick bit of custom CSS to pop the titles and enabling a plugin to adjust the hotkeys, it has satisfactory UI for desktop and mobile.
It isn’t identical to reader, but the selection of features suits me at least as well, and the fact that it is self hosted means I feel like I can rely on it. If any other reader refugees would like an account on my instance to try it, just ask.
One of the most important things of mine google has in Reader has is the searchable index of 7,790 (according to the export tool) starred items that constitute my reference for everything on the internet I gave a shit about in the last several years. This problem is also solved with ttrss – Nicolas Höning’s gritttt has a nice python script for translating the google takeout starred.json into an importable chunk of sql for ttrss. I had a problem the first time I tried to use it, but it turned out to be a malformed json file from that takeout run, not a problem with the script, and things are properly imported now.
The search tools are not quite as responsive as google’s, but are entirely adequate for “I read something about that about $SPAN ago…” tasks, and has a graceful interface for restricting searches to any hierarchical category.
As for other services, I’m still evaluating. Plus is, for the moment, hard to beat as a discussion platform for tech issues, and as an automated way to stash photos from my phone. The former can be replicated with blogs and RSS, and the latter is basically fancy rsync, or any of several third parties, so I think I will keep using those in the low-trust way I have been, particularly since I know all the content is easily exported.
Gmail is unbeatable for handling large volumes of archived email, and the tagging and conversation view features are more developed than any thing else I’ve seen by far, but it is lacking PGP and S/MIME support, and has some serous privacy implications. I’m afraid this is another situation where google has out-competed all the other options. The actual switch of which account is the “real” one is easy, since all my email is already being aliased/routed anyway, and one of the accounts is attached to my hosting. Roundcube still seems to be the best web-mail solution, although Horde (if Bluehost had a less-ancient version) looks pretty attractive too.
As for Android, I’ve never been terribly loyal – it is simply the most open, least broken platform for which I could buy legitimate hardware last time I was shopping. There are some signs that Sailfish, Tizen, Firefox OS, Ubuntu Mobile, or some other random Linux-based mobile OS might actually get a foothold, and that would be fine. For the time being, I’m thinking I’ll start preferentially using secondary markets, especially F-Droid, over the Play store. I already preferentially use FOSS software, I may as well take google out of the loop on it in the process. I still think the lack of line-item veto for permissions in the Android API is possibly the most egregious profit-over-users-well-being decision google has ever made.
So: I’m not wholesale evacuating google products, but my long-established habit of backing all my google data up via takeout (it is a very good thing all around that that exists), and a mail downloader (which has been a little twitchy of late) has been vindicated, and I’ll be watching for alternatives and staying ready to use them, because google has proven counting on their services will end in disappointment.
Trusting the FSF for anything is a bad idea. For truth in reporting is a particularly bad idea.
If you read the jabber-operators thread you will find they did not intentionally break federation. They restricted invites/day from specific domains (they listed three in the thread but this is dynamic probably) and allow most others. The domains they restrict were unusually spammy.
Interesting, I saw some “This explains why users of our corporate XMPP system have been complaining” posts in comment threads places this was reposed, and didn’t see any clear statement on what google actually settled on doing in the jabber-operators thread, so I’ve been assuming it was roughly true.
It does look like the FSF article is characteristically overblown, although the number of times they have been early instead of wrong about that sort of thing makes me give them more credit than they might deserve.