Without ads, Android apps could be more than twice as power-efficient | Hacker News Reader

Without ads, Android apps could be more than twice as power-efficient

theverge.com

84 pointsGreekOphion55 comments

Someone on Reddit bothered to read the actual paper and found the following:

4.4 Accounting for High Rate Components The FSM power model does not cover RAM and Organic LED screen (OLED) since these components are accessed at much higher rates (and hence called high rate components) resulting in high overheads in event based modeling. [...] we leave RAM/OLED accounting as future work.

If I'm understanding that correctly it's a rather big factor to decide to casually ignore.

source: http://www.reddit.com/r/technology/comments/r39lz/free_apps_...

Coming from a mobile ad network myself, we read the paper and there are a few points we'd like to make. Mainly that the study only tested the first 33 seconds of the app launching, which is the shortest session time I can think of, especially for a chess game...

Read the rest: http://www.inmobi.com/devblog/2012/03/21/battery-usage-of-mo...

Why do "journalists" always forget to link to the primary source?

http://research.microsoft.com/en-us/people/mzh/eurosys-2012.... is the original research paper and also provides interesting insights for smartphone app developers.

[edit] Oops, I completely ignored that orange/light grey markup. Must have been an over-active form of ad-blindness.

I see a source link. The Verge has an interesting layout, though; IMO it would be more normal to link inline.

I skipped over that even when I was looking for it on a first pass, assuming it was part of the tagging system.

Linking to a primary source takes people away from your site. That's why news sites don't like to do it. I'd suggest that the amount of inline linking in blogs would be a counterpoint to that argument, but that's another story.

Since when is Microsoft so interested in doing Android-related research?

Since they started developing Windows Phone? It's a little obvious, but why not study the design decisions of your competitors? Especially when it reveals inefficiencies?

And they thought they should just share it with the world? Please. This is just more of Microsoft's FUD tactics to attack competitors in a dirty way.

They were also the ones to pay for that research that uncovered the first Chrome sandbox exploit. I question their motives for all of this.

That's my point. Microsoft has nothing to lose by researching the inefficiencies of their competitors. It's a pretty good business tactic if you have enough money to spare, and since they started developing Windows Phone, it makes even more sense for them to uncover "problems" with Android.

This paper was written by an intern at MSR. It's possible that it was easier to instrument Android or it has more apps or whatever.

You haven't actually read the paper, have you?

No, I've only read articles about it, so I definitely can't make a definite judgment about the quality of the research. But assuming journalists aren't misrepresenting the facts (which may or may not be a good assumption in this situation), it seems fairly obvious that Microsoft has a vested interest in developing a better phone operating system.

Their intentions weren't necessarily nefarious, but I doubt they funded this purely for the sake of science. It's in their best interest to study cell phone efficiency, and it's in their best interest to study their competitor's phones. That just seemed like an obvious conclusion, but is there something in the actual paper that indicates anything different?

So, obviously, the solution here is to have tighter integration of ad platforms with the underlying system, perhaps an ads repo that installed applications can access and which is optimized for power use.

Or, you know, you could charge for your applications and stop putting ads in altogether.

It will be a dark day, my friends, when there is kernel support for advertisements.

the ad services (Google, etc.) doesn't care about your phone's battery life as long as they can serve their ads. They don't get anything in return if they spend time and resources making their ads "more efficient". These companies don't care about the user experience or providing a better product. They're in the ads business not in the making-a-better-phone business.

as far as they're concerned your phone's battery life is a hardware/OS issue.

sad but true.

Google certainly has an interest in extending android battery life, you would be silly to suggest anything else.

All evidence to the contrary (including this article).

All evidence to the contrary?

Android releases since Gingerbread have made much more use of black elements (status bar, menu shortcuts, app launcher background). Bright colours on OLEDs are bad news for battery life.

google is an ad company. the whole reason android exists is to make sure you view as many ads as possible.

you think they don't care about how that affects your battery life? if your phone battery dies, you're not viewing ads anymore. the goal is to have you connected to the internet as much as possible. the more usable their stuff is, the better they can accomplish that goal. the making-a-better-phone business is the ad business for google.

iAd is pretty low level in iOS, right?

The day I see a coca cola logo on my POST screen is the day I purchase a sledgehammer.

If you are not paying for the product, you are paying in other ways, this just puts a number on it. If you consider the implications, you are paying quite a bit in power in addition to privacy issues.

This is why:

a) my gps AND network location is turned off 99.99% of the time.

b) I have ad-blocking software which prevents lots of this network overhead.

^^ the above is justified because it actually costs me, want to serve me ads? fine, but not when they come at a price.

My primary reason to get root on my Android was... well, let's be honest, like every other HN denizen I wanted to explore and control it. But among my other reasons was to be able to run DroidWall, an interface to iptables firewalling.

Same for me. Adaway and LBE followed later.

Thanks for mentioning DroidWall - I can't believe I didn't stumble onto it myself!

> the above is justified because it actually costs me, want to serve me ads? fine, but not when they come at a price

How do you expect them to serve you ads? They have to get to your phone _somehow_ so there will always be network overhead. This just feels like an attempt to justify using applications while removing their revenue stream based on impossible requirements.

How do you expect them to serve you ads? They have to get to your phone _somehow_ so there will always be network overhead.

The amount of network overhead can vary wildly. If the app is polling the ad network API on startup and at a fixed interval, it can keep the radio awake for longer than it would otherwise, or wake it up and keep it in a high power state for the duration of the CELL_DCH timer.

It's not an impossible requirement to queue ads, reduce the frequency of polling, and coalesce the ad network API calls with other app network traffic.

If you are not paying for the product, you are paying in other ways

I have ad-blocking software

Excuse me while I put on my devil's advocate pants... so, you're using the apps without paying, and cheating the developer out of profit?

God yes. Putting ads next to something small and clickable is really irritating. My biggest problem is when there is no paid version of the app without ads. Even after blocking them, I'm still left with a bar of precious, unused screen space.

Is there any adblocker which works without rooting the Android?

Adfree Android.

https://play.google.com/store/apps/details?id=com.bigtincan.....

[i]a) my gps AND network location is turned off 99.99% of the time.[/i]

My understanding was keeping GPS ON does not amount to any battery loss or power usage. GPS is only active for a few seconds when some app request location.

I'm surprised that the ad sevices aren't smarter. Wait until on wifi, cache a bunch of ads, share those ads amongst all the apps that use your service. Is this feasible?

I'm not. There is no incentive for better energy utilization.

[edit] for advertizement driven software. Energy efficiency is something which goes to the bottom line of handset manufacturers. That this research comes from Microsoft, not Google, shows how different the incentives in regards to energy management are for the two companies' mobile OS's.[/edit]

[edit 2]It also illustrates the problem of the commons inherent in the app model. Lots of contact with the ad server is great for each app and poor for overall energy efficiency.[/edit 2]

I mean, there are incentives. If a battery lasts longer, a user will spend more time in apps, which leads to more ad views.

Whether the ad companies are measuring and paying attention to this is a different matter.

They also spend more time in the app when it stutters and lags due to ad load. /snark

Otherwise... it's just another reason to learn smali.

Android has a “where did my battery go” UI to provide that incentive.

It should be. But until there's some competitive pressure for them to do that, they won't.

Feasible in a centralised, OS-level ad system, yes. iAds may do it, I have no idea. But most ad systems are embedded at the app level, so when that app is turned off, no downloading can take place.

You don't need it to be OS-level. When an app with ads turns on, just have it grab a whole bunch of potential ads and cache them in some shared data store (e.g. a Content Provider).

I imagine that the harder part is grabbing the right ads so that you have something relevant to show. There's also some concern about leaking ad information, but I'm not sure how much advertisers care about that.

You don't need it to be OS-level. When an app with ads turns on, just have it grab a whole bunch of potential ads and cache them in some shared data store

That alone would require it to be OS-level, at least on iOS (AFAIK).

You could probably do this with some cleverness using Keychain or a custom URL scheme that an app with ads will only respond to if they have some ads cached. I'm not 100% certain on this though as I've done very little with iOS.

On Android this is obviously much easier. Just use a content provider as I mentioned above.

Well, if only I could tell them that I don't give a shit about dating sites or finding! singles! in! my! area!

At least I do get a relevant ad (for a DirecTV promotion, I don't have cable).

But seeing it 100.000 times is NOT going to make me purchase it.

It seems like it would make daily budgets trickier since not all ads would be shown, and not necessarily on the same day they were served.

This test was run on a Nexus One, which was released in 2010. I'd be curious to see if they got the same results on a newer phone. I'd be willing to bet that this is mostly because ads require more network requests, which aren't as efficient on newer phones.

Something smells of a researcher cherry-picking a result that shows off their tool's results in the best possible light. I suppose I don't have any way to prove that, but I have to ask why the researcher didn't try the test on another phone.

A Nexus One is a very convenient target for research because it is (a) sold without ties to a carrier or any carrier-specific junk, and (b) flashable with any specific system image you want, including custom AOSP builds without doing any other work (the hardware profile for the Nexus One is shipped with AOSP).

With out a screen your phone could run for weeks.

Could someone in ad tech please explain to me: is ad software's inefficiency with power and load times a necessary evil, or is it some gross incompetence from leaders in the field?

I suspect it's neither. It's probably more that advertisers just don't care if their ads use up too much power. Probably the same reason why there are so many annoying Flash ads.

I'm so frustrated that this is on the BBC, on The Verge, it's everywhere. It is TOTAL NONSENSE.

I don't even need to comment much, if you read the original research you will quickly discover just how poor this paper is. They haven't remotely correctly characterised it and a holistic approach would show the ridiculousness of the claims.

People are too busy to read a few pages though, but not too busy to write a bunch of crap and try and get ad impressions.

Can you elaborate on this?

It looks reasonable enough to me. I suppose you could quibble with some of the details of how they assign hardware utilization to software components. Apps with heavy same-resource utilization independent of ad serving might face some mischaracterization, although the basic message (advertising represents a non-negligible power drain) remains intact.

Am I missing something? I admit to having only read the paper once.

If an Android app requires location services and that isn't integral to the functionality of the app, then it's bad design on the dev's part.

It's easy enough to get decent fill rates for advertisements without latex-gloving user privacy and wasting battery life.

Ads are definitely not worth the $2 you save for an app that you like and use regularly. This is just another reason why...

Very few "regular use" apps are advertising-only. All of them come in a "pro" edition for $2 or whatever for the people who care enough to purchase. I like the subtle platform flame, but it's poorly informed and frankly needless. Which is just another reason why...

Reply on news.ycombinator.com