Touch Screen Latency Debate

Ever since Agawi TouchMarks revealed the screen latency test results, a few of the android fans seem to question the methodology itself. From the TouchMark’s blog, “…our latency experts are using their knowledge to introduce the first quantitative and objective benchmark of app response times: TouchMarks. By introducing TouchMarks to the market, we hope to bring more rigour to discussions around touchscreen response times, device lag, streaming latency and other topics related to how responsive an application feels on a mobile device”. Throughout this article, I have mentioned content straight from the iOS devices seem to fare really well in touch responses and hence most users seem to like the interaction with iPhone, iPad, etc. Below graph vindicates the same.

touchmark_graph_1

In addition, Microsoft Research visually (as well as technically) shows us if such issue exist and does it affect any touch device user at all.

These days, people do not believe such results as fanboy element creeps in to most results. Let us see from the comments section of TouchMarks website itself, if there are any valuable points.  First comment from @craigsg said “No one should confuse you for scientists. You have an agenda and preconceived notions” But this reply was quite interesting from @Ian “@craigsj . With regard to perceived latency, we are working on an app where accurate timing is critical to the proper functioning of the app. On iOS we have it working great, on android we are contending with high audio latency + high touch latency. It virtually makes the app not doable on android, without some fairly hacky latency subtractions. We need to be able to measure a touch to within 200ms of an audible cue. With audio latency of 100-400ms and touch latency of >100ms you can see the problem.”

People are good at finding problems in every outcome, I was obviously surprised to hear a comment wherein, the question was about non-selection of stock Android devices. Moreover, visitors felt that, TouchMarks is positively biased towards Apple. @Rohan’s reply was interesting, “Hi Adam, We picked flagship phones from each manufacturer that we had available to us- unfortunately we didn’t have a Nexus 4. The Moto X is pretty close to a stock Android experience though (and it’s actually my personal phone, contrary to most people’s belief I’m an iPhone user now). We did kill the background processes, but didn’t put the phone into airplane mode. It’s an interesting theory though, I’ll test it on a few devices and see if there’s a difference”. People love their devices is not a problem, world is never flat and each individual has different choices and tastes. I would never bother to comment on them. I have been loving Apple devices not because I am a fan of the company, but the products seem to work well every time as per my expectations that includes feel of experience which could vary. I like snappy fluid scrolls but that itself is relative. These researches try to convert subjective remarks to numbers and I am happy to see the effort put by TouchMarks. I guess following remarks best conclude the whole thing:

@Ritesh “Almost all high end Android phones since 2012 have the touch layer fused with the glass. That includes all the Android phones in this test and many more like the One X, GS3, LG OG, Nexus 4 from 2012. You’re probably talking about cheaper/older Android phones. LG and Samsung make the displays/touchscreens for Apple, so it’d be really odd if their own devices didn’t use the same. Hardware is not the issue here. Software is.

I’m an Android user and also an audio engineer+producer. I know first hand how bad latency on Android is. It does not affect daily usage but it definitely sticks out like a thorn when it comes to realtime apps such as midi controllers and synths. For audio apps, I use LivKontrol and TouchDAW on a Nexus 10 and LivKontrol and TouchOSC on an iPad 3. The difference in response is pretty noticeable.

Having said that, Android is getting a lot of work done in the audio latency department and the next version will have major improvements with an option for a realtime kernel. Once that arrives, I’m sure Agawi would do the tests again.”