Project dream

Andrew Dalke adalke at mindspring.com
Sun Dec 28 20:51:42 CET 2003


Me:
> > An integrated chemical/bioinformatics development and
> > exploration environment.  Got a few million dollars to fund me?
> > (Actually, more like 10 million, but bootstrappable with only
> > a few million.)

Will Stuyvesant:
> I guess all that money is needed for getting high quality people into
> it?  Or other reasons?  What techniques would you use?

Have you scoped out how much it costs to run a company?  Suppose
you want to pay a decent salary.  Around here that's about $60 -
$70,000 for software developers, and about the same for good
computational chemists.  (Maybe a bit higher, unlikely to be lower.)
Say 2 software developers, one with good comp. chemistry
experience, one with good QA skills, another for documentation,
and if it's the above environment then a usability engineer and
GUI developer -- call it 4 full-time equivalents, or about $1/4 million
per year.  Add in overhead (health insurance, bookkeeping,
computers, etc.) and that's about $1/2 million per year.

If you've read "Crossing the Chasm", science -- especially the
so-called 'hard sciences' -- are on the early-adopter side of
the chasm.  That is, there are a bunch of people who spent
years in libraries, doing research, writing papers, to get a PhD.
They have few qualms about spending more time and effort
to learn something.  Compare this to sane people, who just
want things to work without usually needing to worry about
the details.  I like that my car, a very complex device, needs
amazingly little attention from me.  Scientists in their specialty
are the grease monkeys like my uncle who complained about
a 1938 model year truck which no longer had the door in
the firewall to tweak the carburetor while driving.  One
chemist, when told by others that the software took a long
time to learn, replied "suck it up and just learn it."

I think this is bad, for two reasons.  First, extending the
analogy a bit, it means the specialist are limited to 1930s
model cars instead of getting to 1960s rally cars; still able
to tweak and tune as needed, but it works and lets you
go places.  Hence my company's motto "More Science,
Less Time."

Second, it prevents non-specialists from using the tools
for that field.  There are many times when someone in
a related field (say, bioinformatics instead of chemical
informatics) needs use a few techniques, or times when
tasks once considered esoteric -- like a similarity search
of a subset of a DNA database -- are taught in
introductory-level courses.  These must be made both
easy to use and robust in operation.

The problem with either approach is that it requires a
lot of new software to be written or rewritten, for reasons
I won't get into.  Scientists tends to be more reactionary
than average, and stay with software where all the warts
are known rather than worrying about new bugs, or
having to justify in talks and papers why a new approach
was used over the standard one.

Putting all these together, that means any good software
for this field will take a lot of work, a lot of time, and a
long, slow acceptance period, which means it will take
3 to 4 years before my project dream is self-supporting.
Hence, "a few million dollars."

Instead, I'm doing consulting -- but still looking for a
company or three interested in funding me to work on
my dream.  :)

> And I can not do C bindings (too long ago I did C).
> Besides I don't *want* to C again :-)

Tools like SWIG make it quite easy.

> > My answer then is to do any of these projects; they are all great
> > ones to learn how to do larger, more useful projects.
>
> Um, more useful?  Can you give some more examples of what
> you think useful?

He-he.  I'm very biased in this respect, so everything I
can think of is in the computational life sciences, and
even when I relax that restriction I still end up with
things that I might make a living out of.  "Useful" at this
point in my life means making money to live on, travel,
enjoy myself -- ie, the proverbial "life, liberty, and the
pursuit of happiness."

The list you gave (two games, a vector graphics program,
and a weblog program) don't fall into that category for
me since there's too much competition already in those
domains.

> I am looking for a fun project and inspiration and maybe even fun
> people who join a project, you got that right :-)

Go to sourceforge and browse the list until you find
something interesting.  If you want, you can even limit
the search to Python.  Or go to bioinformatics.org and
look through the list there. ;)

                    Andrew
                    dalke at dalkescientific.com






More information about the Python-list mailing list