1 |
On Wed, Aug 26, 2020 at 11:33 PM Michał Górny <mgorny@g.o> wrote: |
2 |
|
3 |
> On Wed, 2020-08-26 at 19:39 -0700, Alec Warner wrote: |
4 |
> > If the objective is to compare the cost of the Foundation to an umbrella; |
5 |
> > I'm all for it! |
6 |
> > |
7 |
> > My primary critique of the blog post is in the financials and income |
8 |
> > forecasting. I have two major objections. |
9 |
> > |
10 |
> > (1) The foundation makes more than simply individual donations. The |
11 |
> > methodology where we ignore a large portion of revenue seems rather |
12 |
> > arbitrary to me. |
13 |
> |
14 |
> Any model is 'rather arbitrary' by the very definition of it. I've |
15 |
> never claimed it's perfect or precise. I've explicitly considered |
16 |
> the worst case scenario. |
17 |
> |
18 |
> > I might work on a more numerical model. Really what I |
19 |
> > think we want is some moving average of past years (by revenue source) |
20 |
> and |
21 |
> > we say things like "well we make an average of Y$ from source Z, and we |
22 |
> > made this over a period of A years, so we can forecast some of this |
23 |
> revenue |
24 |
> > into future years." It would be a discounted model, but I don't think |
25 |
> it's |
26 |
> > valuable to discount these extra sources of revenue to 0. Or to put this |
27 |
> > another way; why don't we discount individual paypal donations to 0 also? |
28 |
> > The answer appears to be because we have a historical model that says we |
29 |
> > are likely to get some recurring donation revenue...which then leads me |
30 |
> to |
31 |
> > ask why we are not applying this same heuristic to other revenue sources |
32 |
> > for the Foundation? So in short, I don't agree with only using paypal |
33 |
> > donations and we should forecast other revenue sources. |
34 |
> > |
35 |
> |
36 |
> I've included all sources in the chart to let people decide. The main |
37 |
> thing about 'small donations' is that they're distributed over a lot of |
38 |
> people. So yes, arbitrarily it makes sense to assume that all of people |
39 |
> donating to Gentoo won't suddenly stop doing that. |
40 |
> |
41 |
> The problem with 'big donations' is that they're bound to a single |
42 |
> entity. They're a binary thing. Either we get them, and we have lots |
43 |
> of money, or we don't and we don't. One year a person donates $15000 |
44 |
> and we're rich. Another year only $5000, and our revenue halves. Yet |
45 |
> another year there's no such donation. Yes, surely you could try |
46 |
> averaging it and assuming that *maybe* next year there will be another |
47 |
> $5000 donation and it will even out. Or maybe not because it was just |
48 |
> one person donating. |
49 |
> |
50 |
> Commission and interest are small enough we can skip them for now. |
51 |
> |
52 |
> GSoC money is somewhat constant lately but it all hangs on decision of |
53 |
> one entity. Google says we're in, cool. Google says we're out, |
54 |
> we don't have that money. And it's not exactly that our GSoC project |
55 |
> count is improving. |
56 |
> |
57 |
|
58 |
I understand why you modeled things this way. I think my problem is that |
59 |
when modelling I have two goals. |
60 |
- Be accurate |
61 |
- Be as simple as possible |
62 |
|
63 |
I think your model meets goal 2, but not goal 1, because it wildly |
64 |
underestimates future revenue of Gentoo Foundation Inc and this causes |
65 |
profound problems in accuracy (as I stated earlier.) |
66 |
|
67 |
I'll try to propose an alternative that uses moving averages; so we can |
68 |
better understand. |
69 |
- Recurring paypal revenue |
70 |
- one-off paypal revenue |
71 |
- Large donations |
72 |
- GSoC (minus reimbursements) |
73 |
|
74 |
Again I'm not saying "we will book the same revenue year over year" for |
75 |
these items, but its unlikely we will book 0$ every year[0]. Sadly when I |
76 |
discussed this with 2 accountants they basically said there is no real |
77 |
standard for forecasting revenue. There are common ways to model...but I |
78 |
can't just recommend some forecasting accounting standard because one does |
79 |
not exist to my knowledge :( |
80 |
|
81 |
-A |
82 |
|
83 |
[0] Down years are fine given our cash reserves. At our current burn rate |
84 |
we can likely fund Gentoo for 10 years without earning another cent in |
85 |
income for that entire 10 year period. That is crazy! But it also means we |
86 |
don't need to care too much about annual fluctuations in revenue because we |
87 |
have small expenses and large cash reserves. If our cash reserves go down |
88 |
or our expenses go up, this becomes more of a pressing concern because down |
89 |
years may cause cash flow problems....but this is not a problem for Gentoo |
90 |
at this time. |
91 |
|
92 |
|
93 |
> |
94 |
> -- |
95 |
> Best regards, |
96 |
> Michał Górny |
97 |
> |
98 |
> |