Shashanks Blog: Mar 14, 2008

...the harm was done: the topic became known as
“computer science”---which, actually, is like referring to
surgery as “knife science” --- and it was firmly implanted
in people's minds that computing science is about
machines and their peripheral equipment.

Programming ought not to be an experimental science,
but most working programmers do not have the luxury of
engaging in what Dijkstra means by computing science.
We must work in the realm of experimentation, just as some,
but not all, physicists do. If thirty years from now
programming can be performed without experimentation, it
will be a great accomplishment of Computer Science.

The kinds of experiments you will have to perform include:

Testing systems with small examples to verify that they
conform to the documentation or to understand their response
when there is no documentation,
Testing small code changes to see if they actually fix
a bug,
Measuring the performance of a system under two
different conditions due to imperfect knowledge of there
performance characteristics,
Checking the integrity of data, and
Collecting statistics that may hint at the solution to difficult or hard-to-repeat bugs.

I don't think in this essay I can explain the design of experiments;
you will have to study and practice. However, I can offer two
bits of advice.

First, try to be very clear about your hypothesis, or the assertion
that you are trying to test. It also helps to write the hypothesis
down, especially if you find yourself confused or are working with
others.

You will often find yourself having to design a series of
experiments, each of which is based on the knowledge gained from
the last experiment. Therefore, you should design your experiments
to provide the most information possible. Unfortunately, this
is in tension with keeping each experiment simple---you will
have to develop this judgment through experience.

Team Skills

Why Estimation is Important

To get a working software system in active use as quickly as possible
requires not only planning the development,
but also planning the documentation,
deployment, marketing.
In a commercial project it also requires sales and finance.
Without predictability of the
development time, it is impossible to plan these effectively.

Good estimation provides predictability.
Managers love it,
as well they should. The fact that it is impossible, both
theoretically and practically, to predict accurately how long it
will take to develop software is often lost on managers.
We are
asked to do this impossible thing all the time, and we must face
up to it honestly. However, it would be dishonest not to admit
the impossibility of this task,
and when necessary, explain it. There is a lot of room for
miscommunication about estimates, as people have a startling
tendency to think wishfully that the sentence:

I estimate that, if I really understand the problem, it
is about 50% likely that we will be done in five weeks
(if no one bothers us during that time).

really means:

I promise to have it all done five weeks from now.

This common interpretation problem requires that you
explicitly discuss what the estimate means with
your boss or customer as if they were a simpleton.
Restate your assumptions, no matter how obvious they seem to you.

How to Estimate Programming Time

Estimation takes practice. It also takes labor. It takes so
much labor it may be a good idea to estimate the time it will
take to make the estimate, especially if you are asked to
estimate something big.

When asked to provide an estimate of something big,
the most honest thing to
do is to stall.
Most engineers are enthusiastic and eager
to please, and stalling certainly will displease the stalled.
But an on-the-spot estimate probably won't be accurate and honest.

While stalling, it may be possible to consider doing or prototyping
the task. If political pressure permits, this is the most accurate
way of producing the estimate, and it makes real progress.

When not possible to take the time for
some investigation, you should first establish the meaning of the
estimate very clearly. Restate that meaning as the first and
last part of your written estimate. Prepare a written
estimate by deconstructing the task into progressively smaller
subtasks until each small task is no more than a day;
ideally at most in length.
The most important thing
is not to leave anything out. For instance, documentation,
testing, time for planning, time for communicating with other
groups, and vacation time are all very important. If you spend
part of each day dealing with knuckleheads, put a line item for
that in the estimate. This gives your boss visibility into what
is using up your time at a minimum, and might get you more time.

I know good engineers who pad estimates implicitly,
but I recommend that you do not. One of the results
of padding is trust in you may be depleted. For instance,
an engineer might estimate three days for a task that
she truly thinks will take one day. The engineer may
plan to spend two days documenting it, or two days working
on some other useful project. But it will be detectable
that the task was done in only one day (if it turns out that way),
and the appearance of slacking or overestimating is born.
It's far better to give proper visibility into what you are
actually doing. If documentation takes twice as long
as coding and the estimate says so, tremendous advantage
is gained by making this visible to the manager.

Pad explicitly instead. If a task will probably take one
day---but might take ten days if your approach doesn't
work---note this somehow in the estimate if you can; if not, at
least do an average weighted by your estimates of the probabilities.
Any risk factor that you can identify and assign an
estimate to should go into the schedule. One person is
unlikely to be sick in any given week. But a large
project with many engineers will have some sick time;
likewise vacation time. And what is the probability of
a mandatory company-wide training seminar? If it can
be estimated, stick it in.
There are of course, unknown unknowns, or unk-unks.
Unk-unks by definition cannot be estimated individually.
You can try to create a global line item for all unk-unks,
or handle them in some other way that you communicate to your boss.
You cannot, however, let your boss forget that they exist, and it is
devilishly easy for an estimate to become a schedule without
the unk-unks considered.

In a team environment, you should try to have the
people who will do the work do the estimate, and you
should try to have team-wide consensus on estimates.
People vary widely in skill, experience, preparedness,
and confidence. Calamity strikes when a strong programmer
estimates for herself and then weak programmers are held
to this estimate.
The act of having the whole team agree
on a line-by-line basis to the estimate clarifies the team
understanding, as well as allowing the opportunity for
tactical reassignment of resources (for instance, shifting
burden away from weaker team members to stronger).

If there are big risks that cannot be evaluated, it is your
duty to state so forcefully enough that your manager does
not commit to them and then become embarrassed when the risk occurs.
Hopefully in such a case whatever is needed will be done to decrease the risk.

If you can convince your company to use Extreme Programming, you
will only have to estimate relatively small things, and this is
both more fun and more productive.

How to Find Out Information

The nature of what you need to know determines how you should find it.

If you need information about concrete things that are
objective and easy to verify, for example the latest
patch level of a software product, ask a large number of people
politely by searching the internet for it or by posting on
a discussion group.
Don't search on the internet for anything that smacks
of either opinion or subjective interpretation:
the ratio of drivel to truth is too high.

If you need general knowledge about something subjective the
history of what people have thought about it, go to the library
(the physical building in which books are stored). For example,
to learn about math or mushrooms or mysticism, go to the library.

If you need to know how to do something that is not trivial get two
or three books on the subject and read them. You might learn how
to do something trivial, like install a software package, from the
Internet. You can even learn important things, like good programming
technique, but you can easily spend more time searching and sorting
the results and attempting to divine the authority of the results than
it would take to read the pertinent part of a solid book.

If you need information that no one else could be expected to know
for example, ‘does this software that is brand new work on gigantic
data sets?’, you must still search the internet and the library.
After those options are completely exhausted, you may design
an experiment to ascertain it.

If you want an opinion or a value judgment that takes into account
some unique circumstance, talk to an expert.
For instance, if you want
to know whether or not it
is a good idea to build a modern database management system
in LISP, you should talk to a LISP expert and a database expert.

If you want to know how likely it is that a faster algorithm for
a particular application exists that has not yet been published,
talk to someone working in that field.

If you want to make a personal decision that only you can make
like whether or not you should start a business,
try putting into writing a list of arguments for and against the
idea. If that fails, consider divination.
Suppose you have studied the idea from all angles,
have done all your homework, and worked out all the consequences
and pros and cons in your mind, and yet still remain indecisive.
You now must follow your heart and tell your brain to shut up.
The multitude of available divination techniques are very useful
for determining your own semi-conscious desires, as they each present a complete
ambiguous and random pattern that your own subconscious will assign meaning to.

How to Utilize People as Information Sources

Respect every person's time and balance it against your own.
Asking someone a question accomplishes far more than just
receiving the answer. The person learns about you, both by
enjoying your presence and hearing the particular question.
You learn about the person in the same way, and you may
learn the answer you seek. This is usually far more important than your question.

However, the value of this diminishes the more you do it.
You are, after all, using the most precious commodity
a person has: their time.
The benefits of communication
must be weighed against the costs. Furthermore, the
particular costs and benefits derived differ from person
to person. I strongly believe that an executive of 100
people should spend five minutes a month talking to each person
in her organization, which would be about 5%
of their time. But ten minutes might be too much, and five
minutes is too much if they have one thousand employees.
The amount of time you spend talking to each person in your
organization depends on their role (more than their position).
You should talk to your boss more than your boss's boss,
but you should talk to your boss's boss a little. It may
be uncomfortable, but I believe you have a duty to talk a
little bit to all your superiors, each month, no matter what.

The basic rule is that everyone benefits from talking to you a little bit,
and the more they talk to you, the less benefit they derive.
It is your job to provide them this benefit, and to get the benefit
of communicating with them, keeping the benefit in balance with the time spent.

It is important to respect your own time. If talking to
someone, even if it will cost them time, will save you a
great deal of time, then you should do it unless you think
their time is more valuable than yours, to the tribe, by that factor.

A strange example of this is the summer intern. A summer
intern in a highly technical position can't be expected
to accomplish too much; they can be expected to pester the
hell out of everybody there. So why is this tolerated?
Because the pestered are receiving something important from the intern.
They get a chance to showoff a little. They get a chance to hear
some new ideas, maybe; they get a chance to see things from
a different perspective. They may also be trying to recruit
the intern, but even if this is not the case there is much to gain.

You should ask people for a little bit of their wisdom and
judgment whenever you honestly believe they have something to say.
This flatters them and you will learn something and teach them
something. A good programmer does not often need the advice of
a Vice President of Sales, but if you ever do,
you be sure to ask for it. I once asked to listen in on a few sales
calls to better understand the job of our sales staff.
This took no more than 30 minutes but I think that small effort made
an impression on the sales force.

How to Document Wisely

Life is too short to write crap nobody will read;
if you write crap,
nobody will read it.
Therefore a little good documentation is best.
Managers often don't understand this,
because even bad documentation gives them a false sense of security
that they are not dependent on their programmers.
If someone absolutely insists that you write truly useless
documentation, say ``yes'' and quietly begin looking for a better job.

There's nothing quite as effective as putting an accurate estimate
of the amount of time it will take to produce good documentation
into an estimate to slacken the demand for documentation.
The truth is cold and hard: documentation, like testing,
can take many times longer than developing code.

Writing good documentation is, first of all, good writing.
I suggest you find books on writing, study them, and practice.
But even if you are a lousy writer or have poor command of the
language in which you must document, the Golden Rule is all you really need:
``Do unto others as you would have them do unto you.''
Take time to really think about who will be reading your
documentation, what they need to get out of it, and how you
can teach that to them. If you do that, you will be an above
average documentation writer, and a good programmer.

When it comes to actually documenting code itself, as opposed
to producing documents that can actually be read by non-programmers,
the best programmers I've ever known hold a universal sentiment:
write self-explanatory code and only document code in
the places that you cannot make it clear by writing the code itself.
There are two good reasons for this.
First, anyone who needs to see code-level documentation
will in most cases be able to and prefer to read the code anyway.
Admittedly, this seems easier to the experienced programmer than
to the beginner.
More importantly however, is that the code
and the documentation cannot be inconsistent if there is no documentation.
The source code can at worst be wrong and confusing.
The documentation, if not written perfectly, can lie, and that is a thousand times worse.

This does not make it easier on the responsible programmer.
How does one write self-explanatory code? What does that even mean? It means:

Writing code knowing that someone will have to read it;
Applying the golden rule;
Choosing a solution that is straightforward,
even if you could get by with another solution faster;
Sacrificing small optimizations that obfuscate the code;
Thinking about the reader and spending some of your precious time to make it easier on her; and
Not ever using a function name like ``foo'',``bar'', or ``doIt''!

How to Work with Poor Code

It is very common to have to work with poor quality code
that someone else has written. Don't think too poorly of them,
however, until you have walked in their shoes. They may
have been asked very consciously to get something done
quickly to meet schedule pressure. Regardless, in order to work
with unclear code you must understand it. To understand it
takes learning time, and that time will have to come out
of some schedule, somewhere, and you must insist on it.
To understand it, you will have to read the source code.
You will probably have to experiment with it.

This is a good time to document, even if it is only for yourself,
because the act of trying to document the code will force you
to consider angles you might not have considered, and the
resulting document may be useful. While you're doing this,
consider what it would take to rewrite some or all of the code.
Would it actually save time to rewrite some of it?
Could you trust it better if you rewrote it?
Be careful of arrogance here. If you rewrite it,
it will be easier for you to deal with, but will it
really be easier for the next person who has to read it?
If you rewrite it, what will the test burden be?
Will the need to re-test it outweigh any benefits that might be gained?

In any estimate that you make for work against code you didn't
write, the quality of that code should affect your
perception of the risk of problems and unk-unks.

It is important to remember that abstraction and encapsulation,
two of a programmer's best tools, are particularly applicable
to lousy code. You may not be able to redesign a large block
of code, but if you can add a certain amount of abstraction to
it you can obtain some of the benefits of a good design without
reworking the whole mess. In particular, you can try to wall off
the parts that are particularly bad so that they may be redesigned
independently.

How to Use Source Code Control

Source code control systems let you manage projects effectively.
They're very useful for one person and essential for a group.
They track all changes in different versions so that no code is
ever lost and meaning can be assigned to changes.
One can create throw-away and debugging code with confidence
with a source code control system, since the code you modify
is kept carefully separate from committed, official code that
will be shared with the team or released.

I was late to appreciate the benefits of source code control
systems but now
I wouldn't live without one even on a one-person project.
Generally they are necessary when you have team working on the same code base.
However, they have another great advantage: they encourage thinking about
the code as a growing, organic system.
Since each change is marked as a
new revision with a new name or number, one begins to think of the
software as a visibly progressive series of improvements.
I think this is especially useful for beginners.

A good technique for using a source code control system is to stay
within a few days of being up-to-date at all time.
Code that can't be finished in a few days is checked in,
but in a way that it is inactive and will not be called,
and therefore not create any problems for anybody else.
Committing a mistake that slows down your teammates is
a serious error; it is often taboo.

How to Unit Test

Unit testing, the testing of an individual piece of coded
functionality by the team that wrote it, is a part of coding,
not something different from it. Part of designing
the code is designing how it will be tested.
You should write down a test plan, even if it is
only one sentence. Sometimes the test will be simple:
``Does the button look good?'' Sometimes it will be complex:
``Did this matching algorithm return precisely the correct matches?''

Use assertion checking and test drivers whenever possible.
This not only catches bugs early, but is very useful later
on and lets you eliminate mysteries that you would otherwise have to worry about.

The Extreme Programming developers are writing extensively on
unit testing effectively; I can do no better than to recommend their
writings.

Take Breaks when Stumped

When stumped, take a break. I sometimes meditate
for 15 minutes when stumped and the problem magically
unravels when I come back to it. A night's sleep
sometimes does the same thing on a larger scale.
It's possible that temporarily switching to any
other activity may work.

How to Recognize When to Go Home

Computer programming is an activity that is also a culture.
The unfortunate fact is that it is not a culture that values
mental or physical health very much.
For both cultural/historical reasons (the need to work at
night on unloaded computers, for example) and because of
overwhelming time-to-market pressure and the scarcity of
programmers, computer programmers are traditionally overworked.
I don't think you can trust all the stories you hear, but
I think 60 hours a week is common, and 50 is pretty
much a minimum. This means that often much more than that is required.
This is serious problem for a good programmer, who is responsible not
only for themselves but their teammates as well.
You have to recognize when to
go home, and sometimes when to suggest that other people go home.
There can't be any fixed rules for solving this problem, anymore
than there can be fixed rules for raising a child,
for the same reason---every human being is different.

Beyond 60 hours a week is an extraordinary effort for me, which
I can apply for short periods of time (about one week), and
that is sometimes expected of me.
I don't know if it is fair
to expect 60 hours of work from a person;
I don't even know if 40 is fair.
I am sure, however, that it is stupid to work so much
that you are getting little out of that extra hour you work.
For me personally, that's any more than 60 hours a week.
I personally think a programmer should exercise noblesse oblige
and shoulder a heavy burden.
However, it is not a programmer's
duty to be a patsy.
The sad fact is programmers are often asked to be patsies in order
to put on a show for somebody, for example a manager trying to
impress an executive.
Programmers often succumb to this because they are
eager to please and not very good at saying no.
There are four defenses against this:

Communicate as much as possible with everyone in the company so that
no one can mislead the executives about what is going on,
Learn to estimate and schedule defensively and explicitly and give
everyone visibility into what the schedule is and where it stands,
Learn to say no, and say no as a team when necessary, and
Quit if you have to.

Most programmers are good programmers, and good programmers
want to get a lot done. To do that, they have to manage
their time effectively. There is a certain amount of
mental inertia associated with getting warmed-up to a
problem and deeply involved in it. Many programmers find
they work best when they have long, uninterrupted blocks of
time in which to get warmed-up and concentrate.
However, people must sleep and perform other duties.
Each person needs to find a way to satisfy both their human
rhythm and their work rhythm. Each programmer needs to do
whatever it takes to procure efficient work periods, such
as reserving certain days in which you will attend only the most critical
meetings.

Since I have children, I try to spend evenings with them sometimes.
The rhythm that works best for me is to work a very long day,
sleep in the office or near the office (I have a long commute
from home to work) then go home early enough the next day to spend
time with my children before they go to bed. I am not comfortable
with this, but it is the best compromise I have been able to work out.
Go home if you have a contagious disease. You should go home if
you are thinking suicidal thoughts. You should take a break
or go home if you think homicidal thoughts for more than a
few seconds. You should send someone home if they show
serious mental malfunctioning or signs of mental illness
beyond mild depression. If you are tempted to be
dishonest or deceptive in a way that you normally are not
due to fatigue, you should take a break. Don't use cocaine
or amphetamines to combat fatigue. Don't abuse caffeine.

How to Deal with Difficult People

You will probably have to deal with difficult people.
You may even be a difficult person yourself.
If you are the kind of person who
has a lot of conflicts with coworkers and authority figures,
you should cherish the independence this implies,
but work on your interpersonal skills without sacrificing your intelligence or principles.

This can be very disturbing to some programmers who
have no experience in this sort of thing and whose previous
life experience has taught them patterns of behavior that
are not useful in the workplace.
Difficult people are
often inured to disagreement and they are less affected
by social pressure to compromise than others.
The key is to respect them appropriately,
which is more than you will want to but not as much as they might want.

Programmers have to work together as a team.
When disagreement arises, it must be resolved somehow,
it cannot be ducked for long. Difficult people
are often extremely intelligent and have something very useful to say.
It is critical that you listen and understand the difficult
person without prejudice caused by the person.
A failure to communicate is often the basis of disagreement
but it can sometimes be removed with great patience.
Try to keep this communication cool and cordial, and don't
accept any baits for greater conflict that may be offered.
After a reasonable period of trying to understand, make a decision.

Don't let a bully force you to do something you don't agree with.
If you are the leader, do what you think is best.
Don't make a decision for any personal reasons, and be prepared
to explain the reasons for your decision. If you are a teammate
with a difficult person, don't let the leader's decision have any
personal impact. If it doesn't go your way,
do it the other way whole-heartedly.

Difficult people do change and improve.
I've seen it with my own eyes, but it is very rare.
However, everyone has transitory ups and downs.

One of the challenges that every programmer but especially
leaders face is keeping the difficult person fully engaged.
They are more prone to duck work and resist passively than others.

Personal Skills

How to Stay Motivated

It is a wonderful and surprising fact that programmers are
highly motivated by the desire to create artifacts that are
beautiful, useful, or nifty. This desire is not unique to
programmers nor universal but it is so strong and common among
programmers that it separates them from others in other roles.

This has practical and important consequences.
If programmers are asked to do something that is not
beautiful, useful, or nifty, they will have low morale.
There's a lot of money to be made doing ugly, stupid, and boring stuff;
but in the end, fun will make the most money for the company.

Obviously, there are entire industries organized around motivational
techniques
some of which apply here. The things that are specific to
programming that I can identify are:

Use the best language for the job.
Look for opportunities to apply new techniques,
languages, and technologies.
Try to either learn or teach something, however small, in
each project.

Finally, if possible, measure the impact of your work
in terms of something that will be personally motivating.
For example, when fixing bugs, counting the number of
bugs that I have fixed is not at all motivational to me,
because it is independent of the number that may still
exist, and is also affects the total value I'm adding
to my company's customers in only the smallest possible way.
Relating each bug to a happy customer, however, is
personally motivating to me.

How to be Widely Trusted

To be trusted you must be trustworthy.
You must also be visible.
If know one knows about you, no trust
will be invested in you.
With those close to you, such as your teammates,
this should not be an issue.
You establish trust by being responsive and informative to those
outside your department or team.
Occasionally someone will abuse this trust,
and ask for unreasonable favors.
Don't be afraid of this,
just explain what you would have to give up doing to perform the favor.

Don't pretend to know something that you don't.
With people that are not teammates,
you may have to make a clear distinction
between ``not knowing right off the top of my head''
and ``not being able to figure it out, ever.''

How to Tradeoff Time vs. Space

You can be a good programmer without going to college,
but you can't be a good intermediate programmer without
knowing basic computational complexity theory.
You don't need to know ``big O'' notation, but I personally
think you should be able to understand the difference
between ``constant-time'',``n log n'' and ``n squared''.
You might be able to intuit how to tradeoff time against
space without this knowledge, but in its absence you will not have
a firm basis for communicating with your colleagues.

In designing or understanding an algorithm, the amount of
time it takes to run is sometimes a function of the size
of the input. When that is true, we can say an
algorithm's worst/expected/best-case running time is
``n log n'' if it is proportional to the size ($n$) times
the logarithm of the size. The notation and way of
speaking can be also be applied to the space taken up
by a data structure.

To me, computational complexity theory is beautiful and
as profound as physics---and a little bit goes a long way!

Time (processor cycles) and space (memory) can be
traded off against each other. Engineering is about
compromise, and this is a fine example. It is not always systematic.
In general, however, one can save space by encoding things
more tightly, at the expense of more computation time when
you have to decode them. You can save time by caching,
that is, spending space to store a local copy of something,
at the expense of having to maintain the consistency of the cache.
You can sometimes save time by maintaining more
information in a data structure.
This usually cost a small amount of space but
may complicate the algorithm.

Improving the space/time tradeoff can often
change one or the other dramatically.
However, before you work on this you should ask yourself if
what you are improving is really the thing that needs
the most improvement.
It's fun to work on an algorithm,
but you can't let that blind you to the cold hard fact that improving
something that is not a problem will not make any noticeable difference
and will create a test burden.

Memory on modern computers appears cheap, because unlike processor
time, you can't see it being used until you hit the wall;
but then failure is catastrophic.
There are also other hidden costs to using memory,
such as your effect on other programs that must be resident, and
the time to allocate and deallocate it. Consider this carefully
before you trade away space to gain speed.

How to Stress Test

Stress testing is fun.
At first it appears that the purpose of stress testing
is to find out if the system works under a load.
In reality, it is common that the system does work under
a load but fails to work in some way when the load is heavy enough.
I call this hitting the wall or
bonking^[1].
There may be some exceptions,
but there is almost always a ‘wall’. The purpose of stress testing is
to figure out where the wall is, and then figure out how to move the
wall further out.

A plan for stress testing should be developed early in the project,
because it often helps to clarify exactly what is expected.
Is two seconds for a web page request a miserable failure or a smashing success?
Is 500 concurrent users enough? That, of course, depends,
but one must know the answer when designing the system
that answers the request.
The stress test needs to model reality well enough to be useful.
It isn't really possible to simulate 500 erratic and
unpredictable humans using a system
concurrently very easily, but one can at least create 500 simulations
and try to model some part of what they might do.

In stress testing, start out with a light load and load the system along
some dimension---such as input rate or input size---until you hit the wall.
If the wall is too close to satisfy your needs, figure out which resource
is the bottleneck (there is usually a dominant one.) Is it memory, processor,
I/O, network bandwidth, or data contention? Then figure out how you can move
the wall. Note that moving the wall, that is, increasing the maximum load the
system can handle, might not help or might actually hurt the
performance of a lightly loaded system. Usually performance under
heavy load is more important than performance under a light load.

You may have to get visibility into several different dimensions to
build up a mental model of it; no single technique is sufficient.
For instance, logging often gives a good idea of the wall-clock time
between two events in the system, but unless carefully constructed,
doesn't give visibility into memory utilization or even data structure size.
Similarly, in a modern system, a number of computers and many software
systems may be cooperating. Particularly when you are hitting the wall
(that is, the performance is non-linear in the size of the input)
these other software systems may be a bottleneck.
Visibility into these systems,
even if only measuring the processor load on all participating machines,
can be very helpful.

Knowing where the wall is is essential not only to moving the wall,
but also to providing predictability so that the business can be managed effectively.

How to Balance Brevity and Abstraction

Abstraction is key to programming. You should
carefully choose
how abstract you need to be.
Beginning programmers in their enthusiasm often
create more abstraction than is really useful.
One sign of this is if you create classes that don't
really contain any code and don't really do anything
except serve to abstract something. The attraction of
this is understandable but the value of code brevity must
be measured against the value of abstraction.
Occasionally, one sees a mistake made by enthusiastic idealists:
at the start of the project a lot of classes are defined that
seem wonderfully abstract and one may speculate
that they will handle every eventuality that may arise.
As the project progresses and fatigue sets in,
the code itself becomes messy.
Function bodies become longer than they should be.
The empty classes are a burden to document
that is ignored when under pressure.
The final result would have been better if the energy spent on
abstraction had been spent on keeping things short and simple.
This is a form of speculative programming.
I strongly recommend the article ``Succinctness is Power'' by
Paul Graham[PGSite].

There is a certain dogma associated with useful techniques such as
information hiding and object oriented programming that are
sometimes taken too far. These techniques let one code
abstractly and anticipate change.
I personally think, however, that you should not produce
much speculative code. For example, it is an accepted style
to hide an integer variable on an object behind mutators and accessors,
so that the variable itself is not exposed, only the little interface to it.
This does allow the implementation of that variable to be
changed without affecting the calling code, and is perhaps
appropriate to a library writer who must publish a very stable API.
But I don't think the benefit of this outweighs the cost of the
wordiness of it when my team owns the calling code and hence
can recode the caller as easily as the called.
Four or five extra lines of code is a heavy price
to pay for this speculative benefit.

Portability poses a similar problem.
Should code be portable to a different computer, compiler,
software system or platform, or simply easily ported?
I think a non-portable, short-and-easily-ported piece of code is better than a
long portable one. It is relatively easy and certainly a
good idea to confine non-portable code to designated areas,
such as a class that makes database queries that are specific to a given DBMS.

How to Learn New Skills

Learning new skills, especially non-technical ones, is the greatest
fun of all. Most companies would have better morale if they understood how much this motivates programmers.

Humans learn by doing. Book-reading and class-taking are useful.
But could you have any respect for a programmer who had never
written a program? To learn any skill, you have to put yourself
in a forgiving position where you can exercise that skill.
When learning a new programming language, try to do a small
project it in before you have to do a large project.
When learning to manage a software project, try to manage a small one first.

A good mentor is no replacement for doing things yourself,
but is a lot better than a book. What can you offer a
potential mentor in exchange for their knowledge?
At a minimum, you should offer to study hard so their time won't be wasted.

Try to get your boss to let you have formal training,
but understand that it often not much better than the same amount
of time spent simply playing with the new skill you want to learn.
It is, however, easier to ask for training than playtime in our
imperfect world, even though a lot of formal training is just
sleeping through lectures waiting for the dinner party.

If you lead people, understand how they learn and assist them
by assigning them projects that are the right size and
that exercise skills they are interested in. Don't forget that
the most important skills for a programmer are not the technical ones.
Give your people a chance to play and practice courage,
honesty, and communication.

Learn to Type

Learn to touch-type. This is an intermediate skill because writing code
is so hard that the speed at which you can type is irrelevant and can't
put much of a dent in the time it takes to write code, no matter how
good you are.
However, by the time you are an intermediate programmer you will probably
spend a lot of time writing natural language to your colleagues and others.
This is a fun test of your commitment;
it takes dedicated time that is not much fun
to learn something like that.
Legend has it that when Michael Tiemann^[2]
was at MCC people would stand outside his door to listen
to the hum generated by his keystrokes
which were so rapid as to be indistinguishable.

How to Do Integration Testing

Integration testing is the testing of the integration of
various components that have been unit tested.
Integration is expensive and it comes out in the testing.
You must include time for this in your estimates and your schedule.

Ideally you should organize a project so that there is not
a phase at the end where integration must explicitly take place.
It is far better to gradually integrate things as they are
completed over the course of the project.
If it is unavoidable estimate it carefully.

Communication Languages

There are some languages, that is, formally defined syntactic systems,
that are not programming languages but communication languages---they
are designed specifically to facillitate communication through standardization.
In 2003 the most important of these are UML, XML, and SQL. You should have some
familiarity with all of these so that you can communicate well and decide when
to use them.

UML is a rich formal system for making drawings that describe designs.
It's beauty lines in that is both visual and formal, capable of
conveying a great deal of information if both the author and the audience
know UML. You need to know about
it because designs are sometimes communicated in it. There are very
helpful tools for making UML drawings that look very professional.
In a lot of cases UML is too formal, and I find myself using a simpler
boxes and arrows style for design drawings. But I'm fairly sure UML is
at least as good for you as studying Latin.

XML is a standard for defining new standards. It is not a solution to
data interchange problems, though you sometimes see it
presented as if it was.
Rather, it is a welcome automation of the
most boring part of data interchange, namely, structuring the representation
into a linear sequence and parsing back into a structure. It provides
some nice type- and correctness-checking, though again only a fraction
of what you are likely to need in practicen.

SQL is a very powerful and rich data query and manipulation
language that is not quite a
programming language. It has many variations,
typically quite product-dependent, which are less important than
the standardized core. SQL is the lingua franca of relational databases.
You may or may not work in any field that can benefit from an understanding
of relational databases, but you should have a basic understanding of them
and they syntax and meaning of SQL.

Heavy Tools

As our technological culture progresses, software technology moves from inconceivable,
to research, to new products, to standardized products, to widely available and
inexpensive products. These heavy tools can pull great loads, but can
be intimidating and require a large investment in understanding.
The intermediate programmer has to know how to manage them
and when they should be used or considered.

To my mind right some of the best heavy tools are:

Relational Databases,
Full-text Search Engines,
Math libraries,
OpenGL,
XML parsers, and
Spreadsheets.

How to analyze data

Data analysis is a process in the early stages of software development,
when you examine a business activity and find the requirements to
convert it into a software application. This is a formal definition,
which may lead you to believe that data analysis is an action that you
should better leave to the systems analysts, while you, the programmer,
should focus on coding what somebody else has designed.
If we follow strictly the software engineering paradigm, it may be
correct. Experienced programmers become designers and the sharpest
designers become business analysts, thus being entitled to think about
all the data requirements and give you a well defined task to carry out.
This is not entirely accurate, because data is the core value of every
programming activity. Whatever you do in your programs, you are either
moving around or modifying data. The business analyst is analyzing the
needs in a larger scale, and the software designer is further squeezing
such scale so that, when the problem lands on your desk, it seems that
all you need to do is to apply clever algorithms and start moving
existing data.

Not so.

No matter at which stage you start looking at it, data is the main
concern of a well designed application. If you look closely at how a
business analyst gets the requirements out of the customer?s requests,
you?ll realize that data plays a fundamental role. The analyst creates
so called Data Flow Diagrams, where all data sources are identified and
the flow of information is shaped. Having clearly defined which data
should be part of the system, the designer will shape up the data
sources, in terms of database relations, data exchange protocols, and
file formats, so that the task is ready to be passed down to the
programmer. However, the process is not over yet, because you ? the
programmer ? even after this thorough process of data refinement, are
required to analyze data to perform the task in the best possible way.
The bottom line of your task is the core message of Niklaus Wirth, the
father of several languages. ?Algorithms + Data Structures = Programs.?
There is never an algorithm standing alone, doing something to itself.
Every algorithm is supposed to do something to at least one piece of
data.

Therefore, since algorithms don't spin their wheels in a vacuum, you
need to analyze both the data that somebody else has identified for you
and the data that is necessary to write down your code.
A trivial example will make the matter clearer. You are implementing a
search routine for a library. According to your specifications, the user
can select books by a combination of genre, author, title, publisher,
printing year, and number of pages. The ultimate goal of your routine is
to produce a legal SQL statement to search the back-end database.
Based on these requirements, you have several choices:
check each control in turn, using a "switch" statement, or several "if"
ones;
make an array of data controls, checking each element to see if it is
set;
create (or use) an abstract control object from which inherit all your
specific controls, and connect them to an event-driven engine.
If your requirements include also tuning up the query performance, by
making sure that the items are checked in a specific order, you may
consider using a tree of components to build your SQL statement.
As you can see, the choice of the algorithm depends on the data you
decide to use, or to create. Such decisions can make all the difference
between an efficient algorithm and a disastrous one.
However, efficiency is not the only concern. You may use a dozen named
variables in your code and make it as efficient as it can ever be. But
such a piece of code might not be easily maintainable. Perhaps choosing
an appropriate container for your variables could keep the same speed
and in addition allow your colleagues to understand the code better when
they look at it next year. Furthermore, choosing a well defined data
structure may allow them to extend the functionality of your code
without rewriting it.
In the long run, your choices of data determines how long your code will
survive after you are finished with it.
Let me give you another example, just some more food for thought.
Let's suppose that your task is to find all the words in a dictionary
with more than three anagrams, where an anagram must be another word in
the same dictionary.
If you think of it as a computational task, you will end up with an
endless effort, trying to work out all the combinations of each word and
then comparing it to the other words in the list.
However, if you analyze the data at hand, you'll realize that each word
may be represented by a record containing the word itself and a sorted
array of its letters as ID. Armed with such knowledge, finding anagrams
means just sorting the list on the additional field and picking up the
ones that share the same ID.
The brute force algorithm may take several days to run, while the smart
one is just a matter of a few seconds. Remember this example the next
time you are facing an intractable problem.

Team Skills

How to Manage Development Time

To manage development time, maintain a concise
and up-to-date project plan. A project plan is an estimate,
a schedule, a set of milestones for marking progress, and an
assignment of your team or your own time to each task on the estimate.
It should also include other things you have to remember to do,
such as meeting with the quality assurance people, preparing
documentation, or ordering equipment. If you are on a team,
the project plan should be a consensual agreement, both at
the start and as you go.

The project plan exists to help make decisions, not to show
how organized you are. If the project plan is either
too long or not up-to-date, it will be useless for making decisions.
In reality, these decisions are about individual persons.
The plan and your judgment let you decide if you should
shift tasks from one person to another.
The milestones mark your progress.
If you use a fancy project planning tool, do not be seduced into
creating a Big Design Up Front (BDUF) for the project, but use it
maintain concision and up-to-dateness.

If you miss a milestone, you should take immediate action such
as informing your boss that the scheduled completion of that
project has slipped by that amount. The estimate and schedule
could never have been perfect to begin with; this creates the
illusion that you might be able to make up the days you missed
in the latter part of the project.
You might.
But it is just
as likely that you have underestimated that part as that you
have overestimated it. Therefore the scheduled completion of
the project has already slipped, whether you like it or not.

Make sure you plan includes time for:
internal team meetings,
demos,
documentation,
scheduled periodic activities,
integration testing,
dealing with outsiders,
sickness,
vacations,
maintenance of existing products, and
maintenance of the development environment.
The project plan can serve as a way to give outsiders or
your boss a view into what you or your team is doing.
For this reason it should be short and up-to-date.

How to Manage Third-Party Software Risks

A project often depends on software produced by organizations
that it does not control. There are great risks associated with third
party software that must be recognized by everyone involved.

Never, ever, rest any hopes on vapor.
Vapor is any alleged software that
has been promised but is not yet available.
This is the surest way to go out of business.
It is unwise to be merely skeptical of a software company's promise
to release a certain product with a certain feature at a certain date;
it is far wiser to ignore it completely and forget you ever heard it.
Never let it be written down in any documents used by your company.

If third-party software is not vapor, it is still risky, but at
least it is a risk that can be tackled.
If you are considering using third-party software, you should
devote energy early on to evaluating it.
People might not like to hear that it will take two weeks or
two months to evaluate each of three products for suitability,
but it has to be done as early as possible.
The cost of integrating cannot be accurately estimated
without a proper evaluation.

Understanding the suitability of existing third party software
for a particular purpose is very tribal knowledge.
It is very subjective and generally resides in experts.
You can save a lot of time if you can find those experts.
Often times a project will depend on a third-party software system
so completely that if the integration fails the project will fail.
Express risks like that clearly in writing in the schedule.
Try to have a contingency plan, such as another system that
can be used or the ability to write the
functionality yourself if the risk can't be removed early.
Never let a schedule depend on vapor.

How to Manage Consultants

Use consultants, but don't rely on them. They are wonderful
people and deserve a great deal of respect.
Since they get to see a lot of different projects,
they often know more about specific technologies and
even programming techniques than you will. The best way to use
them is as educators in-house that can teach by example.

However, they usually cannot become part of the team in
the same sense that regular employees are, if only
because you may not have enough time to learn their
strengths and weaknesses. Their financial commitment
is much lower. They can move more easily.
They may have less to gain if the company does well.
Some will be good, some will be average, and some will be bad,
but hopefully your selection of consultants will not be
as careful as your selection of employees, so you will get more bad ones.

If consultants are going to write code, you must review it carefully
as you go along. You cannot get to the end of the a project
with the risk of a large block of code that has not been reviewed.
This is true of all team members, really,
but you will usually have more knowledge of the team members
closer to you.

How to Communicate the Right Amount

Carefully consider the cost of a meeting; it costs its duration
multiplied by the number of participants.
Meetings are sometimes
necessary, but smaller is usually better.
The quality of communication in small meetings is better, and
less time overall is wasted. If any one person is bored at a
meeting take this as a sing, that the meeting should be smaller.

Everything possible should be done to encourage informal communication.
More useful work is done during lunches with colleagues
than during any other time.
It is a shame that more companies do not recognize nor support this fact.

How to Disagree Honestly and Get Away with It

Disagreement is a great opportunity to make a good decision,
but it should be handled delicately.
Hopefully you feel that you have expressed your thoughts
adequately and been heard before the decision is made.
In that case there is nothing more to say, and you should
decide whether you will stand behind the decision even
though you disagree with it. If you can support this
decision even though you disagree, say so.
This shows how valuable you are because you are independent
and are not a yes-man, but respectful of the decision and a team player.

Sometimes a decision that you disagree with will be made
when the decision makers did not have the full benefit of
you opinion.
You should then evaluate whether to
raise the issue on the basis of the benefit to the company
or tribe. If it is a small mistake in your opinion, it may
not be worth reconsidering. If it is a large mistake in you opinion,
then of course you must present an argument.

Usually, this is not a problem. In some stressful circumstances
and with some personality types this can lead to things being
taken personally. For instance, some very good programmers
lack the confidence needed to challenge a decision even when
they have good reason to believe it is wrong.
In the worst of circumstances the decision maker is insecure and
takes it as a personal challenge to their authority.
It is best to remember that in such circumstances people react
with the reptilian part of their brains.
You should present your argument in private, and try to show
how new knowledge changes the basis on which the decision was made.

Whether the decision is reversed or not, you must remember
that you will never be able to say ‘I told you so!’
since the alternate decision was fully explored.

Judgement

How to Tradeoff Quality Against Development Time

Software development is always a compromise between what the
project does and getting the project done.
But you may be asked to tradeoff quality to speed
the deployment of a project in a way that offends your engineering
sensibilities or business sensibilities. For example,
you may be asked to do something that is a poor software
engineering practice and that will lead to a lot of maintenance problems.

If this happens your first responsibility is to inform your team and
to clearly explain the cost of the decrease in quality.
After all, your understanding of
it should be much better than your boss's understanding.
Make it clear what is being lost and what is being gained,
and at what cost the lost ground will be regained in the next cycle.
In this, the visibility provided by a good project plan should be helpful.
If the quality tradeoff affects the quality assurance effort, point
that out (both to your boss and quality assurance people).
If the quality tradeoff will lead to more bugs being
reported after the quality assurance period, point that out.

If she still insists you should try to isolate the shoddiness into
particular components that you can plan to rewrite or
improve in the next cycle.
Explain this to your team so that they can plan for it.

NinjaProgrammer at Slashdot sent in this gem:

Remember that a good design will be resillient against
poor code implementations. If good interfaces and abstractions exist
throughout the code, then the eventual rewrites will be far more
painless. If it is hard to write clear code that is hard to fix,
consider what is wrong with the core design that is causing this.

How to Manage Software System Dependence

Modern software systems tend to depend on a large number of
components that may not be directly under your control.
This increases productivity through synergy and reuse.
However, each component brings with it some problems:

How will you fix bugs in the component?
Does the component restrict you to particular
hardware or software systems?
What will you do if the component fails completely?

It is always best to encapsulate the component in some way
so that it is isolated and so that it can be swapped out.
If the component proves to be completely unworkable, you may be
able to get a different one, but you may have to write your own.
Encapsulation is not portability, but it makes
porting easier, which is almost as good.

Having the source code for a component decreases
the risk by a factor of four.
With source code, you can evaluate it easier, debug it easier,
find workarounds easier, and make fixes easier.
If you make fixes, you should give them to the owner of the component
and get the fixes incorporated into an official release;
otherwise you will uncomfortably have to maintain an unofficial version .

How to Decide if Software is Too Immature

Using software other people wrote is one of the most
effective ways to quickly build a solid system.
It should not be discouraged, but the risks associated
with it must be examined. One of the biggest risks is
the period of bugginess and near inoperability that is
often associated with software before it matures, through
usage, into a usable product.
Before you consider integrating with a software system,
whether created in house or by a third party,
it is very important to consider if it is really
mature enough to be used. Here are ten questions you should ask yourself about it:

Is it vapor? (Promises are very immature).
Is there an accessible body of lore about the software?
Are you the first user?
Is there a strong incentive for continuation?
Has it had a maintenance effort?
Will it survive defection of the current maintainers?
Is there a seasoned alternative at least half as good?
Is it known to your tribe or company?
Is it desirable to your tribe or company?
Can you hire people to work on it even if it is bad?

A little consideration of these criteria demonstrates the great
value of well-established free software and open-source software
in reducing risk to the entrepreneur.

How to Make a Buy vs. Build Decision

An entrepreneurial company or project that is trying to accomplish
something with software has to constantly make so-called
buy vs. build decisions.
This turn of phrase is unfortunate in two ways:
it seems to ignore open-source and free software which
is not necessarily bought.
Even more importantly, it should perhaps be called an
obtain and integrate vs. build here and integrate
decision because the cost of
integration must be considered.
This requires a great combination of business, management,
and engineering savvy.

How well do your needs match those for which it was designed?
What portion of what you buy will you need?
What is the cost of evaluating the integration?
What is the cost of integration?
What is the cost of evaluating the integration?
Will buying increase or decrease long term maintenance costs?
Will building it put
you in a business position you don't want to be in?

You should think twice before building something that is
big enough to serve as the basis for an entire other business.
Such ideas are often proposed by bright and optimistic people that
will have a lot to contribute to your team. If their idea is
compelling, you may wish to change your business plan; but do
not invest in a solution bigger than your own business
without conscious thought.

After considering these questions, you should perhaps prepare
two draft project plans, one for building and one for buying.
This will force you to consider the integration costs.
You should also consider the long term maintenance costs of both solutions.
To estimate the integration costs, you will have to
do a thorough evaluation of the software before you buy it.
If you can't evaluate it, you will assume an unreasonable risk
in buying it and you should decide against buying that particular product.
If there are several buy decisions under consideration,
some energy will have to be spent evaluating each.

How to Grow Professionally

Assume responsibility in excess of your authority.
Play the role that you desire.
Express appreciation for people's contribution to the
success of the larger organization, as well as things
as that help you personally.

If you want to become a team leader, instigate the formation of consensus.
If you want to become a manager, take responsibility for the schedule.
You can usually do this comfortably while working with a leader or a manager,
since this frees them up to take greater responsibility.
If that is too much to try, do it a little at a time.

Evaluate yourself. If you want to become a better programmer,
ask someone you admire how you can become like them.
You can also ask your boss,
who will know less but have a greater impact on your career.

Plan ways to learn new skills, both the trivial technical kind,
like learning a new software system, and the
hard social kind, like writing well,
by integrating them into your work.

How to Evaluate Interviewees

Evaluating potential employees is not given the energy it deserves.
A bad hire, like a bad marriage, is terrible.
A significant portion of everyone's energy should be devoted to recruitment,
though this is rarely done.

There are different interviewing styles.
Some are torturous, designed to put the
candidate under a great deal of stress.
This serves a very valuable purpose of possibly revealing
character flaws and weaknesses under stress.
Candidates are no more honest with interviewers than they are with themselves,
and the human capacity for self-deception is astonishing.

You should, at a minimum, give the candidate the equivalent
of an oral examination on the technical skills for two hours.
With practice, you will be able to quickly cover what they
know and quickly retract from what they don't know to mark out the boundary.
Interviewees will respect this.
I have several times heard interviewees say that the
quality of the examination was one
of their motivations for choosing a company.
Good people want to be hired for their skills, not where they worked last
or what school they went to or some other inessential characteristic.

In doing this, you should also evaluate their ability to learn,
which is far more important than what they know.
You should also watch for the whiff of brimstone that is given off
by difficult people. You may be able to recognize it by comparing
notes after the interview,
but in the heat of the interview it is hard to recognize.
How well people communicate and work with people is more
important than being up on the latest programming language.

A reader has had good luck using a ‘take-home’ test for
interviewees. This has the advantage that can uncover the
interviewee that can present themselves well but can't really
code---and there are many such people. I personally have not
tried this technique, but it sounds sensible.

Finally, interviewing is also a process of selling.
You should be selling your company or project to the candidate.
However, you are talking to a programmer,
so don't try to color the truth.
Start off with the bad stuff,
then finish strong with the good stuff.

How to Know When to Apply Fancy Computer Science

There is a body of knowledge about algorithms,
data structures, mathematics, and
other gee-whiz stuff that most programmers know about but rarely use.
In practice, this wonderful stuff is too complicated
and generally unnecessary.
There is no point in improving an algorithm when most of your time is
spent making inefficient database calls, for instance.
An unfortunate amount of programming consists of getting systems
to talk to each other and using very simple
data structures to build a nice user interface.

When is high technology the appropriate technology?
When should you crack a book to get something other
than a run-of-the-mill algorithm?
It is sometimes useful to do this but it should be evaluated carefully.

The three most important considerations
for the potential computer science technique are:

Is it well encapsulated so that the risk to other systems is
low and the overall increase in complexity and maintenance cost is low?
Is the benefit startling (for example, a factor of two in a
mature system or a factor of ten in a new system)?
Will you be able to test and evaluate it effectively?

If a well-isolated algorithm that uses a slightly fancy algorithm
can decrease hardware cost or increase performance by a factor of
two across an entire system, then it would be criminal not to consider it.
One of the keys to arguing for such an approach is to
show that the risk is really quite low, since the proposed
technology has probably been well studied, the only issue is
the risk of integration. Here a programmer's experience and
judgment can truly synergize with the
fancy technology to make integration easy.

How to Talk to Non-Engineers

Engineers and programmers in particular are generally recognized by
popular culture as being different from other people.
This implies that other people are different from us.
This is worth bearing in mind when communicating with
non-engineers;
you should always understand the audience.

Non-engineers are smart, but not as grounded in creating
technical things as we are.
We make things.
They sell things and handle things and count things and manage things,
but they are not experts on making things.
They are not as good at working together on teams as engineers are
(there are no doubt exceptions.)^[3]
Their social skills are generally as good as or better than
engineers in non-team environments, but their work does
not always demand that they practice the kind of intimate,
precise communication and careful subdivisions of tasks that we do.

Non-engineers may be too eager to please and they may be intimidated by you.
Just like us, they may say ‘yes’ without really meaning it to please you or
because they are a little scared of you, and then not stand behind their words.

Non-programmers can understand technical things but they do not
have the thing that is so hard even for us---technical judgment.
They do understand how technology works,
but they cannot understand why a certain approach would take
three months and another one three days. (After all, programmers
are anecdotally horrible at this kind of estimation as well.)
This represents a great opportunity to synergize with them.

When talking to your team you will, without thinking, use a
sort of shorthand, an abbreviated language that is effective
because you will have much shared experience about technology
in general and your product in particular.
It takes some effort not to use this shorthand with
those that don't have that shared experience,
especially when members of
your own team are present.
This vocabulary create a wall between you and those that
do not share it, and, even worse, wastes their time.

With your team, the basic assumptions and goals do not need to
be restated often, and most conversation focuses on the details.
With outsiders, it must be the other way around.
They may not understand things you take for granted.
Since you take them for granted and don't repeat them,
you can leave a conversation with an outsider thinking
that you understand each other when really there is a large
misunderstanding. You should assume that you will miscommunicate
and watch carefully to find this miscommunication.
Try to get them to summarize or paraphrase what you are saying
to make sure they understand. If you have the opportunity
to meet with them often, spend a little bit of time asking
if you you are communicating effectively, and how you can do it
better. If there is a problem in communication, seek to
alter your own practices before becoming frustrated with theirs.

I love working with non-engineers.
It provides great opportunities to learn and to teach.
You can often lead by example, in terms of the clarity of your communication.
Engineers are trained to bring order out of chaos, to bring clarity
out of confusion, and non-engineers like this about us.
Because we have technical judgment and can usually understand business issues,
we can often find a simple solution to a problem.

Often non-engineers propose solutions that they think will make
it easier on us out of kindness and a desire to do the right thing,
when in fact a much better overall solution exists which can
only be seen by synergizing the outsiders view with your technical judgment.
I personally like Extreme Programming because it addresses this
inefficiency; by marrying the estimation quickly to the idea, it makes
it easier to find the idea that is the best combination of cost and benefit.

^[1]“to hit”

It is our job to do the hard and discern the impossible.
From the point of view of most working programmers,
something is impossible if either it cannot be grown from a
simple system or it cannot be estimated.
By this definition
what is called research is impossible.
A large volume of mere work is hard,
but not necessarily impossible.

The distinction is not facetious because you may very well
be asked to do what is practically impossible,
either from a scientific point of view or a software
engineering point of view.
It then becomes your job to help the entrepreneur find
a reasonable solution which is merely hard and gets most
of what they wanted. A solution is merely hard when it
can be confidently scheduled and the risks are understood.

It is impossible to satisfy a vague requirement, such as
‘Build a system that will compute the most attractive hair
style and color for any person.’ If the requirement can
be made more crisp, it will often become merely hard, such
as ‘Build a system to compute an attractive hair style and
color for a person, allow them to preview it and make changes,
and have the customer satisfaction based on the original
styling be so great that we make a lot of money.’
If there is not crisp definition of success, you will not succeed.

How to Utilize Embedded Languages

Embedding a programming language into a system has
an almost erotic fascination to a programmer.
It is one of the most creative acts that can be performed.
It makes the system tremendously powerful.
It allows you to exercise her most creative and Promethean skills.
It makes the system into your friend.

The best text editors in the world all have embedded languages.
This can be used to the extent that the intended audience can
master the language. Of course, use of the language can be made optional,
as it is in text editors, so that initiates can use it and no one else has to.

I and many other programmers have fallen into the trap
of creating special purpose embedded languages.
I fell into it twice. There already exist many languages
designed specifically to be embedded languages. You should
think twice before creating a new one.

The real question to ask oneself before embedding a language is:
Does this work with or against the culture of my audience?
If you intended audience is exclusively non-programmers,
how will it help?
If your intended audience is exclusively programmers, would
they prefer an applications programmers interface (API)?
And what language will it be? Programmers don't want to learn
a new language that is narrowly used;
but if it meshes with their culture they
will not have to spend much time learning it.
It is a joy to create a new language.
But we should not let that blind us to the needs of the user.
Unless you have some truly original needs and ideas, why not
use an existing language so that you can leverage the
familiarity users already have with it?

Choosing Languages

The solitary programmer that loves his work (a hacker) can
choose the best language for the task. Most working programmers have
very little control of the language they will use. Generally, this
issue is dictated by pointy-haired bosses who are making a political
decision, rather than a technological decision, and lack the courage
to promote an unconventional tool even when they know, often
with firsthand knowledge, that the less accepted tool is best.
In other cases the very real benefit of unity among the team, and to
some extent with a larger community, precludes choice on the part
of the individual. Often managers are driven by the need to be
able to hire programmers with experience in a given language.
No doubt they are serving what they perceive to be the best
interests of the project or company, and must be respected for that.
However, I personally believe this the most wasteful and
erroneous common practice you are likely to encounter.

But of course, things are never one-dimensional. Even if
a core language is mandated and beyond your control, it is often
the case that tools and other programs can and should be written
in a different language. If a language is to be embedded (and
you should always consider it!) the choice of language will depend
a lot on the culture of the users. One should take advantage of this
to serve your company or project by using the best language for the job, and
in so doing make work more interesting.

Programming languages should really be called notations in
that learning one is not at all as difficult as learning a natural
language. To beginners and to some outsiders ``learning a new language''
seems a daunting task; but after you have three under
your belt it's really just a question of becoming familiar with
the available libraries. One tends to think of a large system
that has components in three or four languages as a messy hodgepodge;
but I argue that such a system is in many cases stronger than
a one-language system in several ways:

There is necessarily loose coupling between the components that
are written in different notations (though maybe not clean interfaces),
You can evolve to a new language/platform easily by rewriting
each component individually,
Its possible that some of the modules are actually up-to-date.

Some of these effects may only be psychological; but psychology matters.
In the end the costs of language tyranny outweigh any advantage that
it provides.

Compromising Wisely

How to Fight Schedule Pressure

Time-to-market pressure is the pressure to deliver a good product quickly.
It is good because it reflects a financial reality,
and is healthy up to a point.
Schedule pressure is the pressure to deliver
something faster than it can be delivered and
it is wasteful, unhealthy, and all too common.

Schedule pressure exists for several reasons.
The people who task programmers do not fully
appreciate what a strong work ethic we have and
how much fun it is to be a programmer.
Perhaps because they project their own behavior onto us, they believe
that asking for it sooner will make us work harder to get it there sooner.
This is probably actually true, but the effect is very small,
and the damage is very great. Additionally, they have no
visibility into what it really takes to produce software.
Not being able to see it, and not be able to create it themselves,
the only thing they can do is see time-to-market pressure and
fuss at programmers about it.

The key to fighting schedule pressure is simply to
turn it into time-to-market pressure.
The way to do this to give visibility into the relationship
between the available labor and the product.
Producing an honest, detailed, and most of all,
understandable estimate of all the labor involved
is the best way to do this. It has the added advantage
of allowing good management decisions to be made about
possible functionality tradeoffs.

The key insight that the estimate must make plain is that
labor is an almost incompressible fluid.
You can't pack more into a span of time anymore than you
can pack more water into a container over and above that container's volume.
In a sense, a programmer should never say ‘no’, but rather to say
‘What will you give up to get that thing you want?’
The effect of producing clear estimates will be to increase
the respect for programmers.
This is how other professionals behave.
Programmers' hard work will be visible.
Setting an unrealistic schedule will also be painfully obvious to everyone.
Programmers cannot be hoodwinked.
It is disrespectful and demoralizing to ask them to do something unrealistic.
Extreme Programming amplifies this and builds a process around it;
I hope that every reader will be lucky enough to use it.

How to Understand the User

It is your duty to understand the user, and to help
your boss understand the user.
Because the user is not as intimately involved in the creation of
your product as you are, they behave a little differently:

The user generally makes short pronouncements.
The user has their own job; they will mainly think of small
improvements in your product, not big improvements.
The user can't have a vision that represents the complete body of
your product users.

It is your duty to give them what they really want,
not what they say they want.
It is however, better to propose it to them and get
them to agree that your proposal
is what they really want before you begin, but they may not
have the vision to do this.
Your confidence in your own ideas about this should vary.
You must guard against both arrogance and false modesty in
terms of knowing what the customer really wants.
Programmers are trained to design and create.
Market researchers are trained to figure out what people want.
These two kinds of people, or two modes of thought in the same person,
working harmoniously together give the best chance of
formulating the correct vision.

The more time you spend with users the better you will be able to
understand what will really be successful.
You should try to test your ideas against them as much as you can.
You should eat and drink with them if you can.

Guy Kawasaki[Rules] has emphasized the importance of
watching what your users do in addition to listening to them.

I believe contractors and consultants often have tremendous problems
getting their clients to clarify in their own minds what they really
want.
If you intend to be a consultant, I suggest you choose your clients
based on their clear-headedness as well as their pocketbooks.

How to Get a Promotion

To be promoted to a role, act out that role first.

To get promoted to a title,
find out what is expected of that title and do that.

To get a pay raise, negotiate armed with information.

If you feel like you are past due for a promotion,
talk to your boss about it. Ask them explicitly what you need to
do to get promoted, and try to do it. This sounds trite,
but often times your perception of what you need to do
will differ considerably from your boss's.
Also this will pin your boss down in some ways.

Most programmers probably have an exaggerated sense
of their relative abilities in some ways---after all, we
can't all be in the top 10%!
However, I have seem some people who were seriously unappreciated.
One cannot expect everyone's evaluation to perfectly match
reality at all times, but I think people are generally
moderately fair, with one caveat: you cannot be appreciated without
visibility into your work. Sometimes, do to happenstance or
personal habits, someone will not be noticed much. Working from
home a lot or being geographically separated from your team and
boss makes this especially difficult.

Serving Your Team

How to Develop Talent

Nietschze exaggerated when he said[Stronger]:

What does not destroy me, makes me stronger.

Your greatest responsibility is to your team.
You should know each of them well.
You should stretch your team, but not overburden them.
You should usually talk to them about the way
they are being stretched.
If they buy in to it, they will be well motivated.
On each project, or every other project, try to stretch them in
both a way that they suggest and a way that you think will be good for them.
Stretch them not by giving them more work, but by giving them a new
skill or better yet a new role to play on the team.

You should allow people (including yourself) to fail occasionally and should
plan for some failure in your schedule.
If there is never any failure, there can be no sense of adventure.
If there are not occasional failures, you are not trying hard enough.
When someone fails, you should be as gentle as you can with them
while not treating them as though they had succeeded.

Try to get each team member to buy in and be well motivated.
Ask each of them explicitly what they need to be
well-motivated if they are not.
You may have to leave them dissatisfied,
but you should know what everybody desires.

You can't give up on someone who is intentionally not carrying
their share of the load because of low morale or dissatisfaction
and just let them be slack.
You must try to get them well-motivated and productive.
As long as you have the patience, keep this up.
When your patience is exhausted, fire them.
You cannot allow someone who is intentionally working below their
level to remain on the team, since it is not fair to the team.

Make it clear to the strong members of your team that you think
they are strong by saying so in public.
Praise should be public and criticism private.

The strong members of the team will naturally have more difficult
tasks than the weak members of the team.
This is perfectly natural and nobody will be bothered by it as
long as everyone works hard.

It is an odd fact that is not reflected in salaries that a
good programmer is more productive than 10 bad programmers.
This creates a strange situation. It will often be true that
you could move faster if your weak programmers would just get out of the way.
If you did this you would in fact make more progress in the short term.
However, your tribe would lose some important benefits, namely the training
of the weaker members, the spreading of tribal knowledge, and the ability
to recover from the loss of the strong members.
The strong must be gentle in this regard and consider the
issue from all angles.

You can often give the stronger team members challenging, but carefully delineated, tasks.

How to Choose What to Work On

You balance your personal needs against the needs of the team in choosing
what aspect of a project to work on. You should do what you are best at,
but try to find a way to stretch yourself not by taking on more work but
by exercising a new skill. Leadership and communication skills are more
important than technical skills. If you are very strong, take on the
hardest or riskiest task, and do it as early as possible in the project
to decrease risk.

How to Get the Most From Your Teammates

To get the most from your teammates, develop a good team spirit and
try to keep every individual both personally challenged and personally engaged.

To develop team spirit, corny stuff like logoized clothing and parties
are good,
but not as good as personal respect.
If everyone respects everyone else, nobody will want to let anybody down.
Team spirit is created when people make sacrifices for
the team and think in terms of the good of the team before
their own personal good.
As a leader, you can't ask for more than you give yourself
in this respect.

One of the keys to team leadership is to facilitate consensus so
that everyone has buy in. This occasionally means allowing your teammates to be wrong.
That is, if it does not harm the project too much, you must
let some of your team do things their own way, based on
consensus, even if you believe with great confidence it is
the wrong thing to do.
When this happens, don't agree, simply disagree openly
and accept the consensus.
Don't sound hurt, or like you're being forced into it, simply state that you
disagree but think the consensus of the team is more important.
This will often cause them to backtrack.
Don't insist that they go through with their initial plan if they do backtrack.

If there is an individual who will not consent
after you have discussed the issues from all appropriate sides,
simply assert that you have to make a decision and
that is what your decision is.
If there is a way to judge if your decision will be wrong or
if it is later shown to be wrong, switch as quickly as you can
and recognize the persons who were right.

Ask your team, both as a group and individually, what they think
would create team spirit and make for an effective team.

Praise frequently rather than lavishly. Especially praise those who
disagree with you when they are praiseworthy. Praise in public
and criticize in private; with one exception: sometimes growth or
the correction of a fault can't be praised without drawing embarrassing
attention to the original fault, so that growth should be praised in private.

How to Divide Problems Up

It's fun to take a software project and divide it up into tasks that will be
performed by individuals. This should be done early.
Sometimes managers like to think that an estimate can
be made without consideration of the individuals that will perform the work.
This is impossible since the productivity of individuals varies
so widely. Who has particular knowledge about a component also
constantly changes and can have an order of magnitude effect on performance.

Just as a composer considers the timbre of the instrument
that will play a part or the coach of an athletic team considers the
strengths of each player, the experienced team leader will not usually
be able to separate the division of the project into tasks
from the team members to which they will be assigned.
This is part of the reason that a high-performing team should not be broken up.

There is a certain danger in this given that people will become
bored as they build upon their strengths and never improve their
weaknesses or learn new skills. However, specialization is a
very useful productivity tool when not overused.

How to Handle Boring Tasks

Sometimes it is not possible to avoid boring tasks that
are critical to the success of the company or the project.
These tasks can really hurt the morale of those that
have to do them. The best technique for dealing with
this is to invoke or promote Larry Wall's programmer's virtue of Laziness.
Try to find some way to get the computer to do the task
for you or to help your teammates do this.
Working for a week
on a program to do a task that will take a week to do by hand
has the great advantage of being more educational and sometimes
more repeatable.

If all else fails, apologize to those who have to do the boring
task, but under no circumstances allow them to do it alone.
At a minimum assign a team of two to do the work and promote
healthy teamwork to get the task done.

How to Gather Support for a Project

To gather support for a project, create and communicate a
vision that demonstrates real value to the organization as a whole.
Attempt to let others share in your vision creation.
This gives them a reason to support you and
gives you the benefit of their ideas.
Individually recruit key supporters for your project.
Wherever possible, show, don't tell. If possible, construct a
prototype or a mockup to demonstrate your ideas.
A prototype is always powerful but in software
it is far superior to any written description.

How to Grow a System

The seed of a tree contains the idea of the adult but does not fully
realize the form and potency of the adult. The embryo grows.
It becomes larger. It looks more like the adult and has more of the uses.
Eventually it bears fruit.
Later, it dies and its body feeds other organisms.

We have the luxury of treating software like that.
A bridge is not like that; there is never a baby bridge,
but merely an unfinished bridge. Bridges are a lot simpler than software.

It is good to think of software as growing, because it allows us
to make useful progress before we have a perfect mental image.
We can get feedback from users and use that to correct the growth.
Pruning off weak limbs is healthful.

The programmer must design a finished system that can be delivered and used.
But the advanced programmer must do more.
You must design a growth path that ends in the finished system.
It is your job to take a germ of an idea and build a path that takes
it as smoothly as possible into a useful artifact.

To do this, you must visualize the end result and communicate it
in a way that the engineering team can get excited about.
But you must also communicate to them a path that goes from wherever
they are now to where they want to be with no large leaps.
The tree must stay alive the whole time; it cannot be dead at
one point and resurrected later.

This approach is captured in spiral development.
Milestones that are never too far apart are used to mark progress along the path.
In the ultra-competitive environment of business, it
is best if the milestones can be released and make money as
early as possible, even if they are far away from a well-designed endpoint.
One of the programmer's jobs is to balance the immediate payoff
against future payoff by wisely choosing a growth path expressed in milestones.

The advanced programmer has the triple responsibility of growing software, teams, and persons.

A reader, Rob Hafernik, sent in this comment on this section
that I can do no better than to quote in full:

I think you under-emphasize the importance here. It's not just systems,
but algorithms, user interfaces, data models, and so on. It's absolutely
vital as you work on a large system to have measurable progress toward
intermediate goals. Nothing is as bad as the special horror of getting
down to the end and discovering that the whole thing just isn't going to work
(look at the recent debacle of the Voter News System). I would
even go further and state it as a law of nature: no large, complex
system can be implemented from scratch, it can only be evolved
from a simple system to a complex system in a series of intentional
steps.

To which one can only reply Fiat lux!