Matt Harris <matt@netfire.net>:
Interesting concept, and seems like a good idea. What's the end goal look
like?
Depends on timescale. What I want is for a growing number of skilled
engineers to be able to both (a) work shared-infrastructure problems
full time, and (b) be able to feed themselves and pay rent and
live normal lives while doing it.
If this *really* scales up well, shared infrastructure might become
something like a career path. Work on things of widely perceived value,
get lots of patrons, prosper.
We need to do something like this because currently LBIPs are caught between two problems:
(1) No way to monetize critical services
(2) Altruism doesn't scale well.
Encouraging folks to contribute to specific individuals directly may
be a little more difficult though, compared to, say, getting a legitimate
organization going that provides (likely objectively-determined
merit-based) payouts to the sort of folks you're talking about.
I designed loadsharers out of experience that the centralized model has been
tried and failed. As the FAQ notes:
It turns out that recruiting people who are both competent to run
an organization like that and able to sustain the effort is really
hard.
Also, organizations that handle money have high complexity, overhead,
and management costs. Remittance systems offer us a way to route
around most of those costs. Loadsharers is designed to be the thinnest
possible coordination layer over the remittance systems.
Last but not least, centralization creates single points of failure.
A loose network like Loadsharers should be less vulnerable to
individual incompetence, political capture, corruption, etc.
I have specific instances in mind for all the organizational failure
modes I describe.
Also, I have yet to see any evidence that small central panels of
experts are better at judging merit than a swarm attack on the
evaluation problem in which people choose to fund what they like
and know about. That's called a "market". It works.
I think many of us assume that doing the sort of work you're referring to
will definitely result in the regular receipt of many prestigious,
high-paying job offers.
When that happens, it's actually a problem.
Let's suppose that someone were to judge I've been doing high-quality
work on security-hardened NTP. I get a job offer as a result. Is it
going to be to work on NTP? Nope, you can't monetize NTP, so my employer
will want me to work on something else that generates a profit.
Boom. We lose.
If that's not the case, maybe something else we can
do is to help find full-time employment/funding for folks who contribute
and need it.
What "something else" could be more efficient that putting money into
Loadsharers?
Corporate overhead for an employee is typically 100% of gross salary
or up due to plant costs. When you fund an LBIP through a remittance
service, the service takes a cut that's usually about 5%. You could
buy a lot more infrastructure support with the 95% difference.
Part of what the Loadsharers design is surfing on is the fact that
software developers don't actually need the kind of capital
concentration that the modern corporation is adapted to manage.
We used to, back when computers and communications were expensive,
but that was decades ago now.
So, if your corporation wants to do infrastructure support efficiently,
the most effective thing it can do is earmark some number of K$ per
month for the job, then choose experts from among their employees to
put it into Loadsharers, possibly acting as advisers to attract more
money to the things they can make a case are important.
Hope your ankle's feeling better soon!
Thank you, it seems to be healing nicely.