My thanks to Eva Durant, Victor Milne, and Steve Kurtz for your
comments on my long message about combinatorial optimization and
employment.
Unfortunately your comments make it clear to me that once again I've
failed to convey my message. I seem to be talking to an almost
non-existant audience, the group of people who both understand
technical or mathematical matters AND who can think clearly about
society and how it works. Most technical people are uncomfortable
with discussing "soft" issues like the way society works, while most
people who know something about social problem are uncomfortable with
the math. C.P.Snow's two cultures.
Oh, well, I'll try again. Here goes: -- The first line on my home
page says "Imagine a future world in which it is easy to find a good
job." Perhaps it would have captured my intentions better if it had
said "Imagine a future world in which it is easy FOR EVERYONE to find
a good job."
In a previous message I quoted J.A. Campbell, from a technical book on
computing: "the central problem in computer science [is] minimizing or
avoiding the effects of the combinatorial explosions of possible paths
in a search space." Then I suggested that is also the central problem
in in labour and employment, and throughout the rest of society,
wherever matching occurs.
To me this is an utterly fundamental problem, not unrelated to the
laws of thermodynamics, and must be addressed head-on, to be even
approximately solved. But how to deal with it?
A first step is to quantify the problem, to provide some sort of scale
for measuring or describing the combinatorial problem.
I don't know if any of you have looked at my page about a scale for
rating compatibility, http://www.island.net/~dpwilson/scale.html but
when I talk about compatibility on that page I mean it to include
compatible (or suitable, or appropriate) jobs, not just compatibilities
between people.
That scale is based on the idea of a virtual search-space, and I
assign the base-10 logarithm of the size of that virtual space as a
measure of compatibility or suitability. The key part of this is the
idea that we can provide tools -- social technology -- that can allow
people to find the best 1-in-a-million person or job without the need
to actually meet a million persons or apply for a million jobs.
Part of what I've written about but not posted yet is the way this
scale can also be used to estimate the difficulty of finding jobs for
people with very few skills, or those with some mental or physical
handicaps. For an entirely normal, healthy, and well-adjusted person
with perhaps a high-school education, and a few minor skills, a
1-in-a-thousand match may be good enough, but for others a better
match may be necessary.
An illiterate unskilled person, or a handicapped person, might need a
job which is a 1-in-a-million (level 6) match to their few abilities
and interests, in order to do the work and be happy at it, whereas a
normal (etc.) person might only need a 1-in-a-thousand (level 3)
match. But we should have matching technology that can satisfy the
needs of both, and it should be freely available and easy to use.
I had a couple of responses from people who still think there is a
shortage of jobs, which would make my plan unworkable. I disagree.
Please let me spell out my plan in more detail.
First, and most important, don't focus only on the unemployed. As I
repeatedly say, unemployment is only the tip of the iceberg, the real
problem is the mismatch between people and jobs. There's an old joke
which goes:
Q: "How many people work for your company?"
A: "Oh, about 20 percent of them."
That's hilariously true: most businesses have a few employees who
really do their best, and a lot who just put in time. But I don't
blame the ones who just do the minimum, or say they are just lazy,
instead I blame the system which has led them to a job they are not
really suited for. If these employees seem to have an attitude
problem, it is probably because a job or workplace environment they
hate has made them embittered and robbed them of motivation.
I'm convinced that each and every person could be a model employee in
the right job, and in the right work environment, which means with
the right co-workers. But most people never get anywhere near the
right job.
So, one component of a real solution could be a job-matching system,
which would probably be a computerized system, but instead of just
trying to match the unemployed with job-openings, it should work with
the whole of the workforce, trying to get people out of jobs they
are not suited for and into ones that they would excel at.
A more serious and difficult problem than matching people to jobs
is matching people with co-workers. I say that working with a few
people you like and work well with is a pleasure, no matter what the
job is, and the ideal job would be horrible if it involved working
with people you don't like or can't work with.
Therefore, I say we must focus our attention on the formation of
long-lasting teams of people, rather than on jobs and positions.
Rather than thinking of him-or-herself as Assistent Product Manager,
and seeking to get promoted to Product Manager, a person should find
the team of near-ideal co-workers, and think of him-or-herself as part
of the team of "Bob, George, and Helen", whose responsibilities will
jointly increase as they get more expertise and experience as a team.
We need to facilitate this change of perspective, and to do so we need
to carefully consider the personalities of the people involved.
Job-matching is only a fraction of the problem, we also need
personality-matching, for good team formation.
Now what has all this got to do with unemployment? Well, first of
all, people who are badly mismatched to their jobs and co-workers are
less productive, so the whole economy suffers. If we can get people
moved around into the right jobs with the right co-workers, the
economy will recover, and there will be more work for everybody.
Secondly, the ability to properly match people to jobs and co-workers
would mean that employers would be able to fill all the hidden jobs
that currently exist.
For every job that is advertised there is at least one and probably
several jobs that are not advertised. Many employers could employee
one more person if they could find exactly the right person, but don't
advertise these jobs because they only just barely need someone, and
don't want to go to the trouble and expense of filtering through all
the applications and doing all the interviews that hiring involves,
together with the risk of getting an employee who doesn't meet
expectations.
There is a considerable overhead and associated risk involved in
hiring, and employers often decide it is just not worthwhile to do it.
If we could simplify the process and reduce the expense and risk,
these "hidden jobs" would become real jobs.
Finally, the difficulty of finding jobs in the current system means
that employees are reluctant to quit jobs and some employers are
reluctant to lay off or fire people. This decreases the overall
number of people changing jobs and therefore decreases the number of
job openings.
Even though a person who leaves a job adds to the number of persons
seeking jobs, he or she is also adding to the number of available
openings by vacating one, and a situation in which we have more people
seeking work but equally have more job openings is much better than
one in which we have fewer people seeking work but also fewer
openings.
This may sound paradoxical, but it is true. It is easier to find work
if there are more job openings, even if there are also more people
applying for them. The advantages to the job seeker of an extra
opening more than compensate for the disadvantage caused by another
person competing in the job market.
So, as I tried to say in my original piece on unemployment, better
matching of people to jobs and co-workers would also solve the
unemployment problem. But how do we do it? How do we solve the
matching problem? I think the real answer is to make use of the
social network, the network formed by all human beings linked by
whatever means they communicate, including face-to-face conversation.
If this network had not been fragmented by the huge influx of people
into cities and other phenomena of the modern age, we could depend on
it -- the traditional method of finding a job was always to ask your
family and friends, i.e. your links to the social network.
But the social network is a mess. Given time it might be able to heal
itself, but I think we are almost forced into using computers to help
out. The use of personality-matching methods which I suggested above
to help create good teams in the workplace can also help to knit up
the torn fabric of the social network.
There is another perspective on this, that of the employers.
Employers do not simply hire people to make work, they want efficiency
and productivity, and so they want to hire the best people.
If each business hires the "best" people it can get, and pays them the
smallest amount of money needed to get and keep these people, then it
seems that a society composed of such businesses should be nearly
optimal in one sense: -- the whole society should then be as
productive and efficient as possible.
But I think that's just wrong, for rather mathematical reasons, that
are rarely thought about and rarely understood.
In optimization theory people describe a certain class of algorithms
as "greedy" algorithms, because they take the best choices locally,
without concern for the overall goal. It is true that some problems
can be solved by greedy algorithms, but many can't.
Among the problems that cannot be solved by greedy algorithms is the
assignment problem or bipartite matching problem. I claim a society
that depends on companies hiring what they consider the best available
employees for the money, is a society trying to solve a bipartite
matching problem with a greedy algorithm, and we know that can't be
done.
I hope this makes sense to you. Any comments will be greatly
appreciated!
In place of a summary or conclusion, let me just formulate a few hard
questions, which I think we need to answer by empirical study:
1. How far from an optimum solution is the current matching of
employees to jobs?
2. How much of the unemployment problem is attributable to the
inadequacy of current job-search or job-matching methods to
the solution of this problem?
3. How close to an optimum solution could we get with readily
available methods?
4. How much better would such a near-optimum solution be than any
current situation, in terms of productivity, employee job
satisfaction, and impact on the nation's economy?
Until we have some idea of the answers to these questions, I don't
think we can get very far towards the better future that I imagine.
For more information on this please visit my home page -- its URL
is the bottom line of my signature, below. I am sorry for submitting
such a long message, but it is a complicated problem!
dpw
Douglas P. Wilson [EMAIL PROTECTED]
http://www.island.net/~dpwilson/index.html