MIRI is still trying to decide on the best attack plan. Any thoughts/suggestions would be greatly appreciated. Here's some relevant data that I've collected:
It seems like one of the biggest issues will be making sure donors know what other donors are doing during this time period, to prevent overfilling of matching funds, to make sure the right number of people try to spike the donation box at 5, and to coordinate Tyler's "split up donations" idea during non-peak hours. Maybe there could be a single comment here that is being edited through the day so donors know what the best thing to do at that time is? (preliminary idea)
Thanks for posting this. You beat us to is :)
Those planning on giving $500 or more, if you could email me at malo@intelligence.org that would be greatly appreciated.
For those of us who might not be able to spare $500, but still want to maximize our impact with a couple hundred, will you be posting a finalized strategy at the end for medium sized and small donors?
We are coordinating through a Google Group which anyone can sign up for here. I add new people a few times a day.
I've also set up a Hipchat room (no account required) here: https://www.hipchat.com/g5XbLeBKR
I'll be updating the former with information leading up to the event, and was planning on using that plus the HipChat room to coordinate people on the day of the event.
I suspect our chosen strategy might change during the day depending on the giving patterns of the other orgs.
I didn't know about this google group until right now stubling about it. You should probably make it more visible - a "here" is a bit low-key.
Do we know the smallest donation that's eligible for the $2000 award prize or if there are any other penalties for contributing small increments? Maybe it's worth having some people break up their donations and do a ton of small donations to try to grab some of the $2000 prizes.
If we assume that the minimum donation allowed is $1 and that we will not be maxing out the matching funds, then the opportunity cost of a person doing this strategy instead of donating during the one-to-one matching hours is $1.
I'm not confident about my math here, but...
if p(winning 2000 with one donation) * $2000 > $1, then we should try this strategy Which means that p(winning 2000 with one donation) has to be greater than 1/2000 for it to make sense... right?
So if we think that less than 2000 donations will be made in a certain hour (including all of our donations) then we should have people sit in front of their computers for an hour and make a $1 donation every minute until the total number of donations (including ours) is about 2000?
And if we eliminate the opportunity cost by assuming that we could not just instead plop the money into the matching funds part (if they were maxed out or something) then this seems like the right choice to make anyway and we do not run up against the 1/2000 limit.
Edit: I'm wrong. Mixed up the $2000 and the $150 prizes
I'm not sure, but I don't think we will have access to the number of people who donated for all the other charities. And I suspect that something may be wrong in the math, because that strategy of "donate every minute until 2000 donations occur total" would lead to badly overfilling that hour with donations if, say, 1800 donations were made on behalf of all the other charities.
That math looks like you are calculating the expected value of a raffle ticket randomly awarded to one donor with a value of 2000$.
But instead, the 2000$ is awarded to the charity that received the most donations in one hour. So we just have to donate more times than the second most-donated-to charity.
The opportunity cost bound, with the information that Alexi gave, is 200 donations of 10$. If it ever takes more than 200 donations to get the 2000$, more money could have been earned during the 1-1 dollar matching hour.
So I suspect a good strategy would be to pick a set of off-peak hours where few people are donating, and split up the donations during those times to secure multiple 2000$ prizes with a low(ish) number of donations. Maybe use the success or failure of X number of donations during one off-peak hour to estimate how many donations to do during the next off-peak hour?
Of course, all this assumes that the behavior of the other donors conforms to the normal human diurnal cycle. If they are sufficiently crafty, the multiple charities that have this idea and people willing to wake up at 3 AM may make those hours prohibitively expensive.
I doubt it though. Maybe the European Less Wrong readers could donate during those times so those on the west coast don't have to wake up at terrible hours?
And does anyone want to set up a prediction market to estimate the number of donors for the second-largest charity during the 1-6 window?
EDIT: Assuming we do 100 donations of 10 dollars each per hour for those 5 hours, and no other charity can muster 100 donations per hour.... (If we can get the prize for less than 100 donations in one hour, the expected value is greater than donating during the 2-1 matching hour) it should only take ~$5000 earmarked for that time period to get 5 2000 dollar prizes. That looks doable.
That math looks like you are calculating the expected value of a raffle ticket randomly awarded to one donor with a value of 2000$.
But instead, the 2000$ is awarded to the charity that received the most donations in one hour. So we just have to donate more times than the second most-donated-to charity.
Oh, I must have misread it. I thought it was essentially a raffle.
I mixed it up with this part:
$150 added to a random donation each hour, every hour for 24 hours.
It is possible to schedule donation in advance here: http://svgives.razoo.com/story/Machine-Intelligence-Research-Institute (you can't select the hour though).
This should be the fall-back strategy if you can't make it on the planned time-slot.
As you may know, on May 6, there will be a large one-day price-matching fundraiser for Bay Area Charities.
The relevant details are right here at MIRI's official website.
And this is the webpage to visit to donate.
For those of you who didn't read the two links above, here's the important information.
On May 6, MIRI is participating in Silicon Valley Gives...
Why is this exciting for supporters of MIRI? Many reasons, but here are a few.
- Over $250,000 of matching prizes and funds up for grabs, from sources that normally wouldn't contribute to MIRI:
- Two-to-one dollar match up to $50,000 during the midnight hour.
- $2,000 dollar prize for the nonprofit that has the most individual gift in an hour, every hour, for 24 hours.
- $150 added to a random donation each hour, every hour for 24 hours.
- Dollar for Dollar match up to $35,000 during the 7AM hour, and $50,000 during the noon, 6 PM, and 7 PM hours.
- Local NBC stations, radio stations, businesses, and Bay Area foundations will be promoting the Silicon Valley Day of Giving on May 6th. So if MIRI is making a splash with our fundraising that day, it's possible we'll draw attention from media and by extension new donors.
Making the most of this opportunity will require some cleverness and a lot of coordination. We are going to need all the help we can get...
- If you are interested in supporting MIRI with a large donation during the fundraiser... Get in touch with Malo at malo@intelligence.org
- All MIRI supporters have the potential to make a big impact if we can all work together in a coordinated manner. Sign up below (The sign up sheet is on the MIRI announcement page) to receive updates on our strategy leading up to the event, and updates throughout the fundraiser on the best times to give and promote the event.