[ExI] call to arms against a terrible disease: was RE: Jaw-dropping CWRU Alzheimer's breakthrough?

[Adrian Tymes]

Nice idea, but the numbers in practice tend not to work.

On Fri, Feb 10, 2012 at 10:58 AM, spike <spike66 at att.net> wrote:
> Now imagine an enormous and grimly dedicated army of citizen scientist
> volunteers, a million proles who raise Alzheimer mice in their homes, and
> run the little bastards regularly, perhaps several times daily.

A million unpaid citizen scientists focused on one project, which requires
their time on a regular basis (not just their computers' time), would be
difficult to recruit even for this - especially if you limited
yourself to the US
(which, given regulations, you might have to).  But let us say you had
100,000 - for something like this, that might be doable.

> We have an internet central distribution point which collects the data from
> everywhere, similar in principle to GIMPS and Folding at Home and such.

And yet so very different.  GIMPS and Folding at Home are fully automated,
and as such the quality is far better.  Furthermore, they only require
computer time; the people running this can forget it's even there.  That's
the secret to how they recruit so many people: someone just has to
participate once, and then they're "helping" on a recurring basis by not
doing anything (since these run while the computers idle).

Ask anyone who runs a volunteer organization, how difficult it is to get
recurring labor - even untrained labor, even for the best of causes.

> We
> have some standard food we give the mice, plus some study ingredient.  Could
> be bexarotene, or any oddball thing: some medication left over in the
> cabinet, doesn't matter what it is, but let's get enough volunteers to try a
> bunch of things, doesn't even need to be a medication.

You'd have to organize the logistics of sending these out, and then make
sure they're properly identified in the database.  If "mystery ingredient
10-THP" turns out to be a winner, you'd better be able to say what that
ingredient is, but if no one bothered to record it properly before the
experiment (when, so far as anyone knew yet, it'd do no better than the
millions of others that had been tried), you're SOL.

Worse, the 99+% of ingredients that didn't work also need to be identified,
so people don't waste time retesting the same old thing.  (Sure, lots of
people might volunteer to buy plain old aspirin to test.  After the 1,000th
test shows no effect, 10,000 more people running that test really won't
help at all...especially if that's the only test they can think of, that they'd
be able and willing to run.)

> I can envision a setup which could contain about ten mice, cost about
> a few hundred bucks and perhaps 10 a month in recurring costs.

Easily at least $100/month, including raising and feeding the mice.  For
one thing, what's your cheese budget for the month, if you anticipate
running at least a test a day, possibly several?  For another, what about
shipping around stuff that's worth testing but that no one had lying around
(or sending in to some central place, a mystery compound that needed
identification, so the tests run on it would mean anything)?

That's $10,000,000 in total project budget per month.  Even if most of
that comes out of those 100,000 pockets, quite a bit is still going to
have to come from some central source.

> It could
> communicate with the internet,

It?  The automated system itself?  How does it know which drug it is
testing, or the state of the mouse being tested?  A mouse given an
appetite suppressant would be less likely to go for cheese than an
equal one not given anything, and this has nothing to do with
Alzheimer's.

In general, such experiments need to be run in as identical setups
as possible, to screen out unrelated factors.  With the setup you are
proposing, it does not seem practical to do this.

Which is not to say that something like this couldn't work - just a
few problems with this embodiment.  Maybe you can find a way to
address those challenges?