Developing Great Software

1
Developing Great Software

• Lecture 6 July 5,2001
• Peopleware

2
Once upon a time

• About 10 years ago, somebody told me that I would
  be obsolete in one year. Something called Case
  was going to replace me.
• Needless to say it didnt happen
• Could Programmers be Obsolete in the future?

3
If I was in the fast food business

• Would want the business to
• Squeeze out error. Make the machine run as smooth
  as possible
• Take a hard line about people goofing off
• Treat workers like interchangeable machines
• Do things by the book
• Eliminate experimentation

4
Peopleware says

• Encourage error. Learn from it
• Creative people need to goof off at times.
  Software is a creative process
• Managers that treat people like humans (not
  parts) often reap great productivity
• Being reactive all the time (concentrating on
  immediate) discourages our inventiveness

5
Parkinsons Law

• Writing in 1954, British Author C. Northcote
  Parkinson introduced the notion that work expands
  to fill the time allocated to it.
• In other words, we delay many projects until the
  last minute and then we work feverishly to finish
  them. Even if we start a project early enough, it
  may take all of our time to complete it.
  Regardless, there is a lack of efficiency and
  effectiveness.
• Managers take advantage of this law by setting
  unreasonable delivery dates

6
Seven false hopes of Software Management

• 1. New Methodology
• 2. Silver Bullet Tool
• 3. Technology is passing us by
• 4. Changing Programming Languages
• 5. Backlog will force productivity
• 6. Automate away the developers
• 7. Put more pressure on workers

7
Peopleware says

• 1. There are no Silver Bullets
• 2. Same as answer 1
• 3. The discipline of Software Engineering is
  fairly constant (Requirements, ). The hard part
  is in that arena, not technology
• 4. 5 gain at most
• 5. Its a myth
• 6. Human communication towards solving problems
  cant be automated
• 7. They wont theyll enjoy it less

8
Office Environment

• Is this your environment?

9
Does this also describe your situation?

• I get my best work done in the early morning,
  before anybody else arrives
• In one late evening, I can do two or three days
  worth of work
• The number 1 problem from a survey stated that
  noise was the biggest problem.

10
Peopleware Code Games

• A game designed to answer questions about
  productivity
• 600 Developers from 92 companies participated

11
Individual Graphs

• Count on best people being 2.5 times better than
  median performer
• Count on the half that are better-than-median
  outdoing the other half by 21
• Count on Best people outperforming worst by 101

12
Coding Game Rules

• A pair (from same company) works on a benchmark
  coding and testing task
• Participants work in their own work areas
• Both pair members perform exactly the same work
  (Design, Code , Test)
• They can use langauge of choice

13
Conclusions

• They determined that your pair-mate was
  critical. If he/she did well, you did well. If
  they did bad, you were brought down also.
• The study pointed out that the best performers
  tended to be clustered in the best-performing
  teams

14
Conclusions Continued

• They determined that workplace and performance
  had a high correlation. Look at the graph below.

15
What did they prove?

• Better people tend to gravitate towards better
  workplace or does a better workplace mean better
  people?

16
How Developers spend their time
17
Flow Time

• During single-minded work time, people are
  ideally in a state called flow
• Flow is deep nearly meditative involvement
• Usually you are unaware of time
• General sense of euphoria
• Can be broken by a single interruption

18
What can you do

• Close your door!!!
• Turn off the phone
• Gently remind your cube mates to be quiet
• Find some alternative work times

19
The Right People

• More than anything Great Software requires the
  Right People. The magic formula
• Get The Right People
• Make them Happy so they dont want to leave
• Turn them loose!

20
Getting the Right People

• Make Sure there is a good fit.
• Team interview
• Hold an audition

21
Make them Happy

• Turnover is a huge problem
• Retraining Costs
• Demoralizing effects
• Why do people leave
• People view themselves as independent contracters
• Management treats them like parts

22
Peoplewares view on Methodology

• A Methodology is a general systems theory of how
  a whole class of thought-intensive work ought to
  be conducted.
• Comes in the form of a fat book that specifies
  how things should be done
• Small m methodology is a basic approach that one
  takes to getting a job done.

23
Dont make Methodology an enforcement issue

• Methodologies seek to force convergence through
  statue. There is usually backlash
• Better ways to achieve convergence are
• Training (Show people why something is good)
• Tools Get some automated aids (VSS, )
• Peer Review (Walkthroughs, inspections, )

24
Growing Productive Teams

• A Jelled Team
• Jelled Team is a group of people so strongly knit
  that the whole is greater than the sum of the
  parts
• What not to do
• PeopleWare Pg. 125

25
Signs of a Jelled Team

• T-Shirts
• Low Turnover
• Strong Sense of Identity
• Sense of Eliteness
• Obvious enjoyment in their work
• The Mariners are a great example

26
Ways to make a Crappy team

• Defensive Management
• Bureaucracy
• Physical Seperation
• Fragmentation of Peoples time
• Quality reduction of product
• Phony deadlines
• Clique control