Taking /e/OS for a Test Drive

01 Jul 2024 by cthos
About 9 min

Right, so here we are again, just another step in my journey to find a way to opt out of all the “shove generative AI into everything” trend in my personal life. In my last post on the Apple Intelligence announcement, I ended on a note of “well, I guess I’ll just not upgrade”. That’s a viable strategy for a while (which I intend to make use of for as long as I can), but in the long term it might not be viable.

Look, I’m not totally opposed to running an LLM or something on device, but I want it to be on my terms, not baked into every interaction with the operating system. Apple seems best poised to allow that kind of behavior (which is funny given their track record on customizations), while Google and Microsoft are hell bent on shoving this everywhere in search of the next big thing.

Thus, I spent a week running an experiment running one of the many projects based on the Android Open Source Project (AOSP), the open source part of the Android project sans the Google Extras. I figure the LLM stuff won’t make it into the open source side of stuff maybe ever since Google sees it as a moneymaker (or do they?), so it seems like a reasonably safe bet for a while.

In order to simulate what my life would be like in a world where I try to get away from the big players while still holding a smart phone (a dumbphone or Lightphone are always options), I set some rules for myself.

1. No signing into Google services on the phone. They shall not touch the datas.

2. No signing into Apple services either. It’s not as big a problem, but I do use Apple Music all the time.

3. Stay as close to the OS’s preferences as possible.

That’s pretty much it, but I did a bit of a bonus goal while I was doing this. I paired a CMF Watch Pro and used the CMF Buds Pro as my daily drivers for the same time period. I wanted to go with a bit of a “budget premium” experience to challenge myself on the “do I really need the best stuff”

As an aside, my primary watch is an Apple Watch 6 and I see no reason to upgrade.

The gear list for this experiment is:

• Pixel 6 Pro (Used, from Backmarket)

• CMF Watch Pro (Amazon)

• CMF Buds Pro (Amazon)

There are a lot of different flavors of AOSP projects out there for various needs. Lineage OS is probably the most common one out there, and Graphene OS is the extreme privacy / security focused version, but I went with /e/OS because I was already familiar with it and it offers a very interesting sync capability (which I’ll get to at the end of this post).

/e/OS is itself a Lineage fork, by the by

Now, I could have gone off the board to a Linux based phone distro (and I may give that a go in the future) but that’s a big jump I didn’t want to take - and I wanted to try to preserve the “app ecosystem” that I’m used to. Which is a big advantage of /e/OS: it ships with MicroG (which they maintain) and an app store that can pull APKs from the Google Play store anonymously (thus fulfilling rule #1 but still taking advantage of the Play store ecosystem). Jumping to only using progressive web apps (PWAs) and open source apps was a bit too far of a stretch for my daily driving.

Now that I’d picked an OS, I needed to pick a phone that would both be compatible and ideally wouldn’t create more eWaste in the process. My original plan had been to use an old Galaxy Note 10 that we had lying around, but unfortunately Samsung US phones have locked bootloaders that are a pain to unlock. So, I went on Backmarket and found a Pixel 6 Pro Unlocked, and in good condition. I did this for both price and compatibility reasons, before you set out on this endeavor yourself, you’ll want to read the supported device list for your given OS carefully.

Unlocked from Google is an important distinction here. The ones the Carriers sell may not be able to unlock the bootloader as well, leaving you stuck with stock android.

The phone arrived safely and was in pretty good condition. There were a few scuffs, but nothing a thin screen protector couldn’t hide (sparing me the madness of noticing it all the time).

Alright! Let’s go!

/e/OS also has an “Easy installer” that supports some newer devices, which is very fancy and easy. Unfortunately, a Pixel 6 Pro is not one of the supported devices.

Okay, so this part is very straightforward, but requires a few steps and you need a separate computer to do it. That guide is here. At a high level, the process is:

1. Download 2 files from the /e/OS site

2. Ensure Android SDK tools are installed on your computer.

3. Use adb / fastboot to reboot into recovery mode.

4. Use fastboot to unlock the bootloader.

5. Flash the files from step 1 onto the phone.

6. Reboot and enjoy your new OS.

Okay, we need to take a quick second and talk about unlocking the bootloader for those who are unfamiliar. The bootloader tells your phone how to load Android, and it is locked to an official version of Android so that you can be sure that the OS running on your phone is what you and the manufacturer expect it to be. It’s one of the things that prevent hackers with physical access to the device from doing nefarious things.

Now, unlocking the bootloader is both what allows you to flash a different OS onto the device, but it also breaks that integrity guarantee, meaning if a threat actor grabs your phone and has a few minutes, they can do some shenanigans like pull all the data off of the phone.

Some devices support relocking the bootloader. The Pixel 6 Pro is not one of those devices. This can still be fiddly even if it is supported.

/e/OS takes a defensive step against this by encrypting the data at rest, encrypted with your password (or pin) so that if that were to happen, at least it’s behind an encryption wall. Now, they could also potentially load malware onto it, which then also hijacks your password and conceivably sends it somewhere, but those are the risks we take.

At its core, /e/OS is just the stock Android 13 experience without the google services built in. Instead, it ships with MicroG, which is an open source implementation of most of the Google APIs.

What this means is that most Android apps, even the ones that rely on Google Play services will still work. (SafetyNet also works in some contexts, though I don’t have any apps that are using it). You can control / enable / disable MicroG at your leisure, but it is necessary for push notifications to work. In particular, Push notifications must still register with Google services because that’s just how push notifications have to work - there are no alternative push servers. There are some apps that allow polling for notifications, but if you want them in realtime, this is your only option (that I am aware of). These are still anonymous as documented here.

The default Launcher (Bliss) is a lot like older iOS before the app drawer existed - namely, there is no app drawer. All installed apps just appear on the home screen. Also, you can’t place widgets anywhere except the leftmost screen. I’m not the biggest fan of this experience, but I stuck with it for this experiment. If I were to do this long term, I’m liable to install an alternative launcher (I like Nova).

A stock Android phone without installing apps will not be very useful unless you’re going for a minimalist kind of thing, but that’s not me, I want apps. You do have the option of only installing open source apps, but that’s also not going to get me to a fully functional app (Look, I use Discord a lot).

App Lounge has a very handy feature so that I would not break my first rule: You can use an anonymous Google account to download apps from the Play Store. This requires you to additionally trust the /e/foundation to not tamper with the apps as they’re being installed, but the app is open sourced so it can be audited and it wouldn’t behoove the e foundation to slip in some back doors. Anyway, using one of any number of anonymous accounts they maintain, you can essentially side load apps from the official play store, which means I can install Discord without too much trouble.

The other neat feature they’ve baked in is the “Advanced Privacy” which essentially blocks ad trackers at the OS level and gives you a handy report on every tracker it’s blocked and which app the request originated from. This includes stuff like the usual Firebase analytics, all the way down to some less common ones like Qualtrics.

I’m pretty sure they’re doing this purely on the domain of the request and using a block list to make the determination, which means some stuff is probably slipping by, but it’s pretty fun to see that DoorDash is extremely leaky.

Likewise, the Aurora store gives most apps a privacy score out of 10. You’ll see a lot of them with a 0/10 on the privacy list, because of all the tracking.

So one of the other really neat things about /e/OS that I wanted to mention is their cloud service, which they’ve got an integration directly in the OS which allows you to sync files, notes, contacts, emails and the like as you would if you were heavily in this ecosystem. Murena.io hosts this cloud service, which gives you 1 GB of storage for free and charges a reasonable fee for upgrades.

But we’re not in this to simply trade one cloud provider for another, oh no. The coolest part about the cloud software is that it’s essentially a specialized fork of Nextcloud, and it’s open sourced. That means, theoretically, you can run the whole set of cloud services on a server you control, deeply integrated with your mobile operating system. That’s pretty neat.

Now, you can already do this with vanilla Nextcloud and the various Nextcloud apps, but you’ll lack the built in integration. It’s not a big hassle to just use Nextcloud, but it is really appealing that built-in integration supports private servers.

The other thing that is cool is that murena.io also just works as a Nextcloud server, so you can do things like use the Nextcloud Feeds app to sync up with RSS feeds you put up in murena.io. Or use the Nextcloud files app to sync your files if you really want to. Lots of flexibility there to host or not host what you want.

Now that the experiment is over and I’m back in the corporate embrace of Apple, what did I like about the experience?

Overall, I was really impressed at how smoothly everything went… for the most part. I could install all the apps I needed, and the PWA support for other major apps (like Starbucks) meant I didn’t need to install as many apps as I have before on other Android phones.

I was also blissfully free of the “TRY GEMINI!” pop-ups that have started appearing on other Android phones (like my Motorola G5 Power) or whatever Samsung is calling theirs (which also just appeared on my wife’s Galaxy S22).

I was concerned that I wouldn’t be able to install an eSIM and get it working, but I was also pleasantly surprised that worked without any issues as well.

The only major issue I ran into whilst doing this experiment is that I make heavy use of Apple pay on the Apple watch, specifically to store my transit card. Tapping and getting on transit is an essential part of my day, which was sorely lacking on the /e/OS build. Payments are probably never going to work on the phone directly (because of how payment industry infrastructure works and requires direct partnerships).

There are some workarounds. I decided to see if I could get my old Galaxy Watch Active 2 to pair and verify a card via Samsung Pay on the watch, and I’m happy to report that after some wrangling, I got it working. The only major hurdle is that I had to install the Edge browser and make it the default so that Samsung would let me log into the Galaxy Wear app.

Now, this is a Tizen based watch, not a WearOS watch, so I have no idea how a WearOS watch is going to function, but reports on the forums don’t look very promising. If I wind up trying it out, I’ll write a follow up post.

The forums have had some success with Garmin Watch / Garmin Pay, but the banks that Garmin Pay supports are pretty limited in the US.

Either way, I got the Watch Active 2 to verify a card, and I was able to tap and pay at a Petco shortly after verifying it. So, uh, success!

Overall, I think this was a pretty successful experiment, and should I be unhappy with my ability to disable OpenAI integration in iOS 18 I can at least switch my phone if I need to - but given that I can just stay on 17 for an indefinite period of time I’m unlikely to make the jump just yet.

I'll update this post when I've got the review of the CMF watch up with a link to it!

rants tech blog