Update: The takju mystery is solved: it's the U.S. GSM Nexus that Google released this morning. It's pretty clear that this device will be one of the first, if not the very first, to see the Jelly Bean OTA update.

We deal with rumors and potential fakes on a daily basis, and I have to warn you right away - on a scale of unconfirmed to solid, the source of this post is leaning to the left.

Everything you're about to see can be easily spoofed, but the circumstances in which the evidence was allegedly obtained, combined with information I dug up in our web server log and the web make quite a compelling case.

In essence, in this case it's too much fun to speculate and too interesting to keep it away from public analysis and, who knows, maybe even anonymous tips (hint, hint).

The Rumor

This particular rumor was started by Yuku Sugianto, an Android developer whose report I am inclined to trust. Yuku is relatively established in the Play Store (at least 800k installs), an engineer at Found (from what I can tell, he started the project, so he may be the sole founder), and has a history of submitting bugs to the Android team.

While he could be trolling, it's not likely. However, I have to remind you again that everything here may end up being false.

Yuku's report along with the crash log follows:

Android JellyBean (no version number) spotted! I got an automatic crash report from one of my application and surprisingly the device info states JellyBean as the OS version. (Android is known to use no version number for unreleased version, e.g. Honeycomb instead of 11 or 3.0 when it was still developer preview)

From ACRA:
PHONE_MODEL=Galaxy Nexus

MODEL=Galaxy Nexus
PRODUCT=takju (what's this takju thing?)
SERIAL=0149xxxxxxxxxxxx (removed by me)
TIME=1334275846000 (Fri Apr 13 00:10:46 UTC 2012)

Another interesting thing is the location of the external storage, which is in /storage/sdcard0 instead of /mnt/sdcard.

The Interesting Bits

Wow, what a treat (if true). Let's enumerate the interesting parts:

  1. JellyBean, which has long been rumored to be the codename for the next version of Android.
  2. takju build of the Galaxy Nexus - we've seen mysid (Verizon CDMA/LTE), countless variants of yakju (GSM), but never a takju. What could it be?
    One of our theories is it's the unreleased 32GB GSM Galaxy Nexus that was canned for unknown reasons after getting announced earlier this year.
    Another, much less plausible, is that it's the Sprint Nexus. I personally doubt it's the latter because Sprint's version is CDMA/LTE, but it sounds way too close to the GSM one. If anything, Sprint's Nexus would be a variant of mysid (mysids, perhaps?).

Rumor Analysis

Now, before we begin, I have to point out that everything in the crash log, which was collected by ACRA from a remote user's phone during a crash in one of Yuku's applications, can be faked.

However, if the data was indeed tempered with, it was done in a very proficient and believable way by someone who really knew what he or she was doing, taking care of every subtlety. The circumstantial nature of the way Yuku acquired the log (a random crash) makes it much less likely that the info was faked and sent over on purpose, leading me to believe that what we're seeing is indeed legitimate.

Let me dive deeper:

  1. While JellyBean is present in the log, the Android version number isn't. This may seem suspicious at first, but it's actually a common practice at Google to assign a codename without a version during development. The same thing happened in our Ice Cream Sandwich leak.
  2. takju.
    1. As I mentioned, the takju variant of the Galaxy Nexus has never been released, yet it appears to exist somewhere out there. For example, this GLBenchmark page, which contains all variants of the Galaxy Nexus it's ever seen, along with version numbers, lists the following:


      There are also other pages, like this one, but the list is pretty short.

    2. I reached out to an engineer on the Android team (who will remain unnamed), seeking basic info on what takju might be, but got turned away with a "no comment." It could have been a generic statement, but this engineer has been helpful in the past, and if takju never existed, I would have expected the engineer to say that he or she simply didn't know what it was. "No comment" responses are always suspicious.
    3. My bet is on the unreleased 32GB Galaxy Nexus. I can't think of another variant that would be so close, yet not that close, to the current yakju variants.
  3. JRN12B, the build.
    1. This build version looks legitimate for a few reasons. It starts with a J, which JellyBean builds need to start with according to guidelines. The exact build number (or it could be an internal version number or revision control's version - in fact, if I didn't know any better and the Android team didn't use git, I'd say it's an SVN revision number) is 330279, which is sufficiently higher than the builds we saw in IMM76B. Looks like JB is well underway, and if each number increment is an individual commit, it already contains over 35,000 of them. Not bad.
    2. I dug into our web server logs to see if I can spot a pattern and found the following from March:

      Dalvik/1.6.0 (Linux; U; Android JellyBean Build/JRM79)

      JRN12B seems like a future iteration of JRM79, which works nicely with our theory here. Notice the presence of JellyBean in that user-agent as well. Of course, both could be fake.

  4. PRIMELC03, the bootloader. The current 4.0.4 Galaxy Nexus's bootloader is PRIMELA03. Once again, PRIMELC03 checks out and looks like a likely successor.
  5. As Yuku mentions, getExternalStorageDirectory() seems to return /storage/sdcard0 instead of something like /mnt/sdcard or /mnt/sdcard-ext, which could point to new interesting developments in the way JellyBean organizes filesystem mounts. If you're not a developer, you probably won't care, but it's definitely something to look out for in the future.
  6. The only thing that did strike me as weird at first (thanks to jcase for pointing it out) is IS_DEBUGGABLE=false. If the device is signed with dev-keys, why is IS_DEBUGGABLE set to false? I then thought about it some more and realized that the IS_DEBUGGABLE flag refers to the way the app was compiled (<application android:debuggable="false" ... />) and has nothing to do with the OS itself, so it's of no interest to us.


JellyBean, another Galaxy Nexus variant we've never seen before, build numbers - none of it is essential or particularly groundbreaking, but all of it is definitely fun stuff for us to soak in, process, analyze, and dream about at night (you do dream of Android, don't you?).

Did I mention a heavy rumor advisory with a chance of bullshit (I hope you were wearing your skepticism +10 cape)?

All of the above is based on speculations at the moment, which I mentioned numerous times, but after analyzing everything presented, I'm inclined to think that it is indeed legitimate build info from a device that is either in the making or is the Android team's equivalent of one of those 2-headed unicorns pooping rainbows that you can only find in government facilities deep underground. It's a freak, it will not see the light of day, but might as well use it if it's sitting around the office, right?

Thanks, Yuku!

P.S. So, what do you think of the above? Did I miss anything else? Will any anonymous tipsters step up and add their 2 cents into the discussion?