# fee feedback

Straw poll:

Old fee schedule:
\$0, \$5, \$10, \$30, \$90, \$270, \$810, …
New fee schedule:
\$0, \$10, \$30, \$80, \$240, \$730, …

Nerdery:
The old fee schedule has the property that the sum of the previous
values is always exactly half the current value.
The new fee schedule is just 3^i, rounded to the nearest \$10, and
has the above property asymptotically.

But the real question is whether it’s ok for that first amount to be

What do you think? Reply to just me with “\$5 is better” or “\$10 would
be fine” and I’ll compile the responses, or reply-all if you have
other commentary.

Thanks!

On Fri, Sep 2, 2011 at 01:51, Daniel Reeves dreeves@umich.edu wrote:

As an aside, it turns out it’s not very painful to pay hundreds of
dollars to one’s own startup. I’ve lost \$110 on my own yellow brick
roads so far. I should probably go back to selling my contracts to
friends, or do things like pledge the money to whoever catches me off
my road, a la blog.bmndr.com/rails . The rest of you shouldn’t have
that problem though, in theory.

So, here’s the cost of going off your yellow brick road the ith time:

f(i) = \$0 if i=1
f(i) = \$5 if i=2
f(i) = f(2)23^(i-3) otherwise.

Robbie Clarken rightly asks, why not just use 3^i?
Well, I was a little fixated on the constraint that whatever amount
you currently have at risk, the sum of the previous amounts you lost
is half that much.
The above formula is the only way to achieve that.
But if we just went with f(i) = 3^i then the ratio of previous total
lost to current amount at risk would start at 1/3 and asymptotically
approach 1/2.
That’s arguably even better and certainly makes for an elegant formula.
Of course, we need f(1) = \$0 so some inelegance has to creep in.
I’m thinking 3^i but rounded to the nearest \$10: f(i) = 10*round(3^i/10).

That gives this:

f(1) = \$0
f(2) = \$10
f(3) = \$30
f(4) = \$80
f(5) = \$240
f(6) = \$730
f(7) = \$2,190
f(8) = \$6,560
f(9) = \$19,680
f(10) = \$59,050
f(11) = \$177,150
f(12) = \$531,440
f(13) = more than it’s possible to charge to any credit card

Obviously unless you have a major equity stake in Beeminder you won’t
ever be willing to get very deep into that fee schedule.
But the choice of f(2) matters a lot.
Do you all have opinions about whether the first amount at risk should
be \$5 or \$10 or something else?

I’m getting the impression that the real hurdle is not the amount but
putting in your credit card at all.
Some of that is uncertainty about our credit card processing so we
I’m not sure how quickly we’ll do that though so I’d like to convince
at least you harder core fans in the meantime that our current credit
card processing is very safe! I added a little more to the FAQ at
Your credit card info never touches our own servers but goes straight
to stripe.com, which, yes, you’ve never heard of, but I’ve gotten to
know them and they’re amazing and I trust them. They’re a ycombinator
company and they really know their stuff.

If that doesn’t convince you and you’d rather stick with
The more of you who say that, the quicker we’ll add at least one of
those options.

http://dreev.es – search://"Daniel Reeves"

http://dreev.es – search://"Daniel Reeves"

\$10 would be fine - \$5 to me is basically nothing
On Thursday, December 22, 2011 12:14:01 pm Daniel Reeves wrote:

Straw poll:

Old fee schedule:
\$0, \$5, \$10, \$30, \$90, \$270, \$810, …
New fee schedule:
\$0, \$10, \$30, \$80, \$240, \$730, …

Nerdery:
The old fee schedule has the property that the sum of the previous
values is always exactly half the current value.
The new fee schedule is just 3^i, rounded to the nearest \$10, and
has the above property asymptotically.

But the real question is whether it’s ok for that first amount to be

What do you think? Reply to just me with “\$5 is better” or “\$10 would
be fine” and I’ll compile the responses, or reply-all if you have
other commentary.

Thanks!

On Fri, Sep 2, 2011 at 01:51, Daniel Reeves dreeves@umich.edu wrote:

As an aside, it turns out it’s not very painful to pay hundreds of
dollars to one’s own startup. I’ve lost \$110 on my own yellow brick
roads so far. I should probably go back to selling my contracts to
friends, or do things like pledge the money to whoever catches me off
my road, a la blog.bmndr.com/rails . The rest of you shouldn’t have
that problem though, in theory.

So, here’s the cost of going off your yellow brick road the ith time:

f(i) = \$0 if i=1
f(i) = \$5 if i=2
f(i) = f(2)23^(i-3) otherwise.

Robbie Clarken rightly asks, why not just use 3^i?
Well, I was a little fixated on the constraint that whatever amount
you currently have at risk, the sum of the previous amounts you lost
is half that much.
The above formula is the only way to achieve that.
But if we just went with f(i) = 3^i then the ratio of previous total
lost to current amount at risk would start at 1/3 and asymptotically
approach 1/2.
That’s arguably even better and certainly makes for an elegant formula.
Of course, we need f(1) = \$0 so some inelegance has to creep in.
I’m thinking 3^i but rounded to the nearest \$10: f(i) =
10*round(3^i/10).

That gives this:

f(1) = \$0
f(2) = \$10
f(3) = \$30
f(4) = \$80
f(5) = \$240
f(6) = \$730
f(7) = \$2,190
f(8) = \$6,560
f(9) = \$19,680
f(10) = \$59,050
f(11) = \$177,150
f(12) = \$531,440
f(13) = more than it’s possible to charge to any credit card

Obviously unless you have a major equity stake in Beeminder you won’t
ever be willing to get very deep into that fee schedule.
But the choice of f(2) matters a lot.
Do you all have opinions about whether the first amount at risk should
be \$5 or \$10 or something else?

I’m getting the impression that the real hurdle is not the amount but
putting in your credit card at all.
Some of that is uncertainty about our credit card processing so we
I’m not sure how quickly we’ll do that though so I’d like to convince
at least you harder core fans in the meantime that our current credit
card processing is very safe! I added a little more to the FAQ at
Your credit card info never touches our own servers but goes straight
to stripe.com, which, yes, you’ve never heard of, but I’ve gotten to
know them and they’re amazing and I trust them. They’re a ycombinator
company and they really know their stuff.

If that doesn’t convince you and you’d rather stick with
The more of you who say that, the quicker we’ll add at least one of
those options.

http://dreev.es – search://"Daniel Reeves"

Results are in: out of 8 responses only 1 person definitely prefers
the fee schedule to start at \$5.
A couple people pointed out that starting the pledge amount at \$10
would probably save them money, since \$5 is too little to care about
losing.

Maybe we should be applying more empirically rigorous criteria than
the mathematical elegance of the formula…

Thanks for the input, everyone; really appreciated!

On Thu, Dec 22, 2011 at 13:14, Daniel Reeves dreeves@umich.edu wrote:

Straw poll:

Old fee schedule:
\$0, \$5, \$10, \$30, \$90, \$270, \$810, …
New fee schedule:
\$0, \$10, \$30, \$80, \$240, \$730, …

Nerdery:
The old fee schedule has the property that the sum of the previous
values is always exactly half the current value.
The new fee schedule is just 3^i, rounded to the nearest \$10, and
has the above property asymptotically.

But the real question is whether it’s ok for that first amount to be

What do you think? Reply to just me with “\$5 is better” or “\$10 would
be fine” and I’ll compile the responses, or reply-all if you have
other commentary.

Thanks!

On Fri, Sep 2, 2011 at 01:51, Daniel Reeves dreeves@umich.edu wrote:

As an aside, it turns out it’s not very painful to pay hundreds of
dollars to one’s own startup. I’ve lost \$110 on my own yellow brick
roads so far. I should probably go back to selling my contracts to
friends, or do things like pledge the money to whoever catches me off
my road, a la blog.bmndr.com/rails . The rest of you shouldn’t have
that problem though, in theory.

So, here’s the cost of going off your yellow brick road the ith time:

f(i) = \$0 if i=1
f(i) = \$5 if i=2
f(i) = f(2)23^(i-3) otherwise.

Robbie Clarken rightly asks, why not just use 3^i?
Well, I was a little fixated on the constraint that whatever amount
you currently have at risk, the sum of the previous amounts you lost
is half that much.
The above formula is the only way to achieve that.
But if we just went with f(i) = 3^i then the ratio of previous total
lost to current amount at risk would start at 1/3 and asymptotically
approach 1/2.
That’s arguably even better and certainly makes for an elegant formula.
Of course, we need f(1) = \$0 so some inelegance has to creep in.
I’m thinking 3^i but rounded to the nearest \$10: f(i) = 10*round(3^i/10).

That gives this:

f(1) = \$0
f(2) = \$10
f(3) = \$30
f(4) = \$80
f(5) = \$240
f(6) = \$730
f(7) = \$2,190
f(8) = \$6,560
f(9) = \$19,680
f(10) = \$59,050
f(11) = \$177,150
f(12) = \$531,440
f(13) = more than it’s possible to charge to any credit card

Obviously unless you have a major equity stake in Beeminder you won’t
ever be willing to get very deep into that fee schedule.
But the choice of f(2) matters a lot.
Do you all have opinions about whether the first amount at risk should
be \$5 or \$10 or something else?

I’m getting the impression that the real hurdle is not the amount but
putting in your credit card at all.
Some of that is uncertainty about our credit card processing so we
I’m not sure how quickly we’ll do that though so I’d like to convince
at least you harder core fans in the meantime that our current credit
card processing is very safe! I added a little more to the FAQ at
Your credit card info never touches our own servers but goes straight
to stripe.com, which, yes, you’ve never heard of, but I’ve gotten to
know them and they’re amazing and I trust them. They’re a ycombinator
company and they really know their stuff.

If that doesn’t convince you and you’d rather stick with
The more of you who say that, the quicker we’ll add at least one of
those options.

http://dreev.es – search://“Daniel Reeves”