What I want in a SttartUp Opportunity, Mobile Hacking-MackIT

The startup environment or landscape is changing. It seems that the direct route of developing and idea and finding VCs is the idiot option as theodds for failure are quite high. Whereas getting in through direct admission to an incubator or getting in through indirect means is or might be a better way.

Since some helpfull folks at Chicago HackerNews have pointed me towards this direction through some passing my email around I should probably state what I want or envision.

My visions is a mobile app hacker space called MackIT. Where the foucs is on the mobile app ideas, getting seed capital, getting users, etc. But, before that happens I need to collect some resources. New mobile devlopement skills is no problem but I need to pick up some other resources.

Thus, the intermeditate steps are to get some cash projects through one of the incubators in Chicago and than use those good project results to develop a new Chicago Mobile A-team. At first start some hack tranining sessions on mobile programming. 

 

Android EMMA Correction

Goolge has seem to get the EMMA coverage wrong for the last 5 adt releases. The corrections to EMMA in the build.xml are here in my gist:

 

 

How I interview with a Startup

I keep getting cluess mf'ers like weebly founders so this is an explanation of how I choose what startups that I will apply to or choose to do an interview with:

1. I fully examine all conversations including email for these ffactors:

           -Am I told the product inthe very first sentence.

          -do they attempt to get me as a user in 2nd sentence.

 

Why? Because if I do not believe in your product I will not interview or work for your startup. GVIE A SHIT ABOUT THE PRODUCT AND ItS USERS! At least show that caring in a well crafted sales ptich. and no recruiters are terrible at writing sales ptiches that get users and potential employees. 

2. 2nd paragraph should be referring to specific stuff in my resume that matches the context of the positions not filled or proijects that I have contributed to. Not hype words, not generalities. For example most UI/UX web deisgners do not know that in native mobile on java side we do not have intrfaces and hus donot have MVC patterns to uncouple things and thus cannot tightly coupl an applicaiton through TDD testing. Most recrutiers do not know either, hence my choice not to talk to any recruiter as they are not qualified to judge android devloper skills.

If you are the one writing that email and do not know how to do this than use that VC money and get someone that does.

 

3. 3rd paragraph should be explainng user growthcurves. If I have to hunt for it than I immediately wonder what you are hidding and find out what you are hidding. 

Let me give you an example, weebly at 6 years at 4.25 million hits amonth is not anything considering that they shoould in fact have 40 million hits per month at year 6. and the fact that they grew a little less than 1 million in 12 months time is not very good to say the least.

Do not be an idiot like weebly founders, if you know your user growth sucks be upfront about it and do not attempt hood-winking as its a combination of factors together that motivate me to interview or not not one by itself.

Last, I do not interview with recruiters or startups that use recruiters. Nor do I interview with startups that are not in the Greater ChicogLand area as it does not make eocnomic sense for me to move to the highest cost cities in North Amrerica betting on a web startup.

 

CodeDroid Progress

Well, I finally got EMMA code coverage workign after tracking down the Google errors inteh build script. Now the fun part.

In the Android SDK we have 2 tools to functional test, one is MonkeyRunner and one is JavaMonkey(ChimpChat). While they have functional testing interaction they lack the viasual testing facilities but at the same time have the ability to install the main parent application and the instrumented application and the testing application.

In other words one could use Sikuli to bridge the unit testing and cuntional testing worlds and have one set of tests run that collects the unit tests and theri reports, the functional tests and their reports, and collect screenshots of the application all into one nice test report package.

Thus, now its add some extensions to Sikuli that streamline this process.

Build A Tipping Point

Many solutions exist to offer agile code quality in mobile applications. On the continuous server side there are even solutions in the cloud. But there is a danger there.

The danger is do they understand theuse cases and the implications of solving those use cases with the appraoch they have developed and are they willing to share that understanding with the end customer. Startup Mobile Application prototypes is very special use case.

Currenlty, now there are some gems of tools in place in the android ecosystem that surpass the code quality and mobile application quality within the iphone dev ecosystem and those tools due to their open source natrue can be extended.

Firms like design agencies are still in the dark ages as far as understanding that we are dealign with a 4-degree space of application quality rather than a direct design to quality realtionship. By developing these underlying code/gui design tools more we can than control the destiny of a developer service firm that serves these well funded startups that need to do fast mobile application development build-out cyce.

Currenlty, the high amounts of venture capital flowing through the start-up incubator systems indicate that this may be a high demand market to attack as far as supplying a developer/cloud service concentrating on fast agile android mobile application prototype services for a quality price offered to funded-startups.

The other concern is that incubators are not offering the mobile developer who also has the server stack skills the quality co-founder offerings once the incubator matures. This may be a way to side-step that trap and get someem quality stock options at co-founder rewards.

Android Lint via ANT

So that is what the finished HTML Android Lint report looks like without error or warnings. The specific ant script stuff is:

Later on its adding my own android checks to lint beyod the default 80. For right now its get the rest of my codeqa reports to look similar to the Android Lint HTML report.

Mobile Application Quality

One aspect of a collection of hard sciences such as my Biochemistry studies at Purdue University is that we have all this indirect measurements and patterns when collected together forming enough basis for direct conclusions. Let me explain.

You might not remember this from high school, but our physics model of the atom and its components is a non physical mathematical model as he cannot directly measure the atom or often see it. In more understandable language we took a whole bunch of indirect measurements and those patterns were collected into a cohesive theory of a model of an atom.

Mobile Application Quality has the same charateristics in that we have no direct measure of such a concept. But, we have this indirect data that can be analyzed to integrate some pattens together in a model of what mobile application quality should look like.

We are not dealing so much with mobile device-OS fragmention as it still exists on all platforms including iPhone but instead dealing with not having all this indirect measuring of items in one place to examine and analyze patterns. Let me give you an example.

Effectiveness of a User Interface can be said to be measured by the mobile application reviews before the UI refactorring and after the UI refactorring along with the individual measuring of user interactions within items of the UI during the refactoring process. But those measures are stored in different formats and locations. The app reviews are inthe developer Google Play or Apple ITunes dev console while any user interactio data we collect is in the Google Analytics Dev console.

Thus, the general idea on the coding/developer side of things was or is two or three-fold. One, measure some code qaulity things. Two, collect those measures in some way and Three generate some analysis.

It is not a magic bullet where we can point to one report and sa aha, change this. But, if we can associate all these measures with code commit changes per commit than we get close to having direct indications of which patterns to pay attention to. For example, we may see an upswing in high marks on app reviews that when coupled to an increase in downloads can be indirectly associated with changes either in code quality of the application and or changes in the user interface.

Its a different developer infrastructure. But, by empowering the developer in making it easy to collect this indirect measures of different quality items we can empower the developer to make a better mobile application. How do I know this works?

Yeah, you are right as I have no direct measures yet. But, if you look at the traditional IT-shop or design agencies app reviews for the mobile applications they have completed for clients you start to see this pattern of not well designed mobile applications, a lot of user complaints all across the board as far as mobile application quality. One of reasons why I notice is that when someone request me to consider itnerviewing with them I look at the mobile applicaiton end user reviews in iTune or Google Play. I mean why not as its free data on whether that firm produces quality stuff, right?

This is why I want to do things differently. Not just in a different developer workflow on both iphone and the android development side but also in the approach so that we empower the different development layers to produce a high quality mobile application. To get there you have to deal with the fragmentation of quality measures being in different silos and that role is best met by coming up with or developing a new set of developer workflow tools to collate those quality measures.

So I am releasing this week some small steps towards this set of goals in the form of developer tools on the android side and hopefully soon some developer tools on the iphone developer side. 

 

 

UI testing via Instrumented

Guess who supports my theories, well at least indirectly, on UI testing via instrumented testing on Android?  No, PiviotalLabs is still poisoning java devs with their TDD bunk. Would yo belive iPhone Devs?

Yeah, it seems that several iphone devs got in their head that instrumented unit testing was an ideal way to test UIs in mobile iphone application development. And, it seems even some neat libraries were ported to ObjectiveC.

That tells me several things at once. One, it tells me that the debating point that Android application sales are failing due to direct makreting effort failure sinlcuding the implemntation of Google Play are somewhat false. False in that marketing failures only account directly for the less than 40% of the total effort. 

The key quality that helps or feeds into the marketing effort is the quality of the application as verified or if you will quaranteed by the effective UI testing that the coding team has implemented and used. which means that my instrumented combination of TDD and BDD mix is somehat the right approach to in both the short term andlong term in addressing the full marketing issues of mobile applications despite no direct marketing plan or implementation.

This is also why you have ot be very carefull when every blog, etc start echoing a set of tending stories as its usually wrong because blogs never attempt to do their own reseach into anything. 

©SoMe - Tales From A Social Media Expert

I have known the person named LorenFeldman as a spectuator from afar watching from the outside as an artist finds his way through the internet cesspool and tries to find some human relevant meaning to all the insanity. He really understands satire and how to transform that understanding of satire into a video of pure thinking theater.

SoMe is a kickstarter project that he started to tell a human story, a story of hopes of humanity, fragility of humanity, and the future of humanity using Social Media as the canvas. Its a project that you should consider supporting. Hpefully, he reaches his goal and we get to see this SoMe creation as a finshed product.

Android and ECJ

My instrumented agile ant build-set tools for android java applications first release will be published in the next day or two. I have found some interesting tidbits.

One of those tidbits is that when using ECJ as the compiler with ant you need to use componentdef ant time so that you can ue the full javac stuff without major changes. The other tidbit when doing that you need to in your Eclipse IDE turn off automatic builds as otherwise one tries to clober the other and you get dx warnings.

The other tidbits you will be able to read and see in the the comments to the ant build scripts. The probable project name will be Humanoid at my bitbucket.org account. First version uses apiviz for javadocs, will not fully release the doclava '2nd' version until I have more time to figure out how to do embedded docs using doclava as I have other stuff that I want to build and get out(publish).