0:00:05.040,0:00:11.600
Programmer by day tester by night - what do I 
really mean by that? Well, having looked at the

0:00:11.600,0:00:16.880
presentations that came before me, I think there 
is a very logical connection to what some of the

0:00:16.880,0:00:22.880
people that are working on testing we're seeing 
like Maurício and Manuel, but also to what Igor

0:00:22.880,0:00:31.520
was just saying about estimation, so let's dive 
into the kitchen, because, actually, well if my

0:00:31.520,0:00:40.800
slides would move... I actually like to use the 
kitchen as a parallel universe to the programming

0:00:40.800,0:00:47.680
universe that we're typically working in as 
developers. So personally, I really love cooking,

0:00:47.680,0:00:56.480
and I like to make intricate dishes and - and put 
them on the table, but as you can imagine, this

0:00:56.480,0:01:03.040
part of the work in the kitchen I don't typically 
like that much. And I think a lot of people would

0:01:03.040,0:01:09.680
agree with me, that this is really what you try 
to delay as much as possible, and well maybe you

0:01:09.680,0:01:16.000
have a dishwasher, and it's not too bad, but, well 
it's definitely not my favorite part of the job.

0:01:17.600,0:01:23.680
So if I were to ask you, how much time does 
making this beautiful dish really cost,

0:01:23.680,0:01:29.600
and maybe somebody would say 30 minutes, and if I 
were to ask how much time you would spend on doing

0:01:29.600,0:01:37.840
the dishes, that would probably be considerably 
longer than it takes you to cook. So from that

0:01:37.840,0:01:42.240
point of view, there is a message coming in so 
i'm just going to see whether that's something

0:01:43.040,0:01:51.440
for me... No... So from that perspective, 
we've been looking at how people perceive

0:01:51.440,0:01:57.840
the time taken to do tasks that you enjoy, 
versus tasks that you might not enjoy so

0:01:57.840,0:02:02.560
much. There's actually a term for that that's 
chronoception but also it's - it's kind of the

0:02:02.560,0:02:06.320
shortened form of the chrono-perception 
the time it takes to do something.

0:02:08.960,0:02:16.240
And what we were really wondering in the 
studies that we've done is, how do people look

0:02:16.240,0:02:26.880
at software testing, and how much time are they 
spending on software testing. So my slides move -

0:02:26.880,0:02:33.200
we actually did - we created a tool a plug-in 
for some popular development environments. We

0:02:33.200,0:02:40.800
unleashed that on 2442 developers from many 
countries, but the key thing is that by having

0:02:40.800,0:02:48.240
this tool installed for several months, we were 
able to collect 150 person-years of development

0:02:48.240,0:02:53.920
during the study. So this gives us a broad basis 
to really look at how people are doing testing.

0:02:55.120,0:03:00.720
Now, during the installation we had this 
fancy wizard and we were asking people,

0:03:00.720,0:03:05.360
do you do testing, do you do that in the 
IDE, do you do other types of testing,

0:03:06.480,0:03:12.560
any other opinions that people might have on 
testing. So one of the key questions was how

0:03:12.560,0:03:18.240
much time do you think you're spending time 
on engineering test code versus engineering

0:03:18.240,0:03:26.640
what we call production code in the IDE. And the 
answers - the average answer over all the 2443

0:03:27.360,0:03:34.000
developers that installed the tool was 
about 50-50. So 49.5 percent of their time

0:03:34.720,0:03:39.920
was spent on engineering test code and 
50.5 percent was spent on engineering

0:03:39.920,0:03:44.800
production code. Now there is something really 
strange going on here, because there is this

0:03:45.520,0:03:52.480
really famous book - The Mythical Man Month from 
Frederick Brooks - and in that book it's written

0:03:52.480,0:03:58.400
that testing takes approximately 50 percent of 
our time. So it's actually very strange that the

0:03:58.400,0:04:06.640
book from - that was published at the end of the 
1970s - actually premonitioned the results of this

0:04:06.640,0:04:13.840
first question in our study. So maybe there is 
actually some folklore going on in our profession.

0:04:14.480,0:04:21.360
Now the key question obviously is that after 
observing these developers, after minimally five

0:04:21.360,0:04:27.200
months and potentially even two years or more, 
what would be the answer? How much time would

0:04:27.200,0:04:33.280
actually be spent on engineering test code? That 
is dramatically different from the estimations.

0:04:34.160,0:04:41.040
One quarter of their time - of these developers 
- was spent on engineering test code in the IDE

0:04:41.040,0:04:44.960
and 75 percent was spent on 
engineering the production code.

0:04:46.320,0:04:53.600
Why is that? Well, these remain assumptions, 
but the fact that software testing is really a

0:04:53.600,0:04:58.560
destructive activity, while engineering production 
code is actually something that you bring to life,

0:04:58.560,0:05:02.640
something that you probably enjoy more, just 
like the cooking versus doing the dishes,

0:05:03.200,0:05:09.840
could be a reason here why people 
overestimate their testing efforts.

0:05:11.600,0:05:17.520
But there is more to this study because we also 
looked at how people do test driven development,

0:05:17.520,0:05:23.920
and this really touches upon what Davide was 
telling a few presentations before. So I think

0:05:23.920,0:05:29.360
we all know what test driven development is - it's 
the core idea that you write your tests first and

0:05:29.360,0:05:36.880
then implement parts of the production code so 
that your tests pass. And this really requires

0:05:36.880,0:05:43.120
this cycle where you have a failing test, you make 
it pass, your refactor, and you continue doing

0:05:43.120,0:05:50.800
that. So from the 574 developers, which is like 
just under 20 percent of the developers in our

0:05:50.800,0:05:58.080
study, that claimed to do TDD, there were actually 
only 47 that really did TDD. Now you can say,

0:05:58.080,0:06:03.280
well, maybe they did it outside of the observation 
- observation period, that could all well be,

0:06:03.280,0:06:08.960
but I think the key message here is that a lot 
of people think they do it, but in practice

0:06:08.960,0:06:14.640
they don't do it that often. Because even for 
those developers that we did observe that they

0:06:14.640,0:06:21.200
were doing TDD they only did it in very specific 
cases, probably because it's not that easy to do.

0:06:23.440,0:06:30.720
So what does that really bring us to? Well, 
the first takeaway is what developers think

0:06:30.720,0:06:38.080
they do or they estimate their efforts on, is 
typically not what they really do in practice. So

0:06:38.080,0:06:43.200
the chronoception of software engineering tasks 
is really something that we need to be careful

0:06:43.200,0:06:53.920
for. And secondly, maybe a plea is that we need to 
understand software engineering tasks better: how

0:06:53.920,0:07:01.840
we do it, and also the effort that it takes to do 
them. And we need to have better tools to estimate

0:07:02.400,0:07:05.680
how much time does it take to test, 
how much time does it really take to

0:07:05.680,0:07:12.560
implement the feature. So my plea here is really 
to also use software analytics during developments

0:07:12.560,0:07:18.720
to better understand what you and your team are 
doing and also create more reliable estimations.

0:07:20.960,0:07:26.880
So finally the two takeaways of my talk: 
the chrono perception - take into account

0:07:26.880,0:07:33.200
that certain things take more or less time 
depending on certain factors in your personality.

0:07:33.840,0:07:43.840
And a general plea to do more software 
analytics to become a better programmer. Thanks.