Does crowdmatching make sense?

Hello,

i would like to have more clarity on the following questions:

  • Will crowdmatching actually work?
  • Does crowdmatching make sense?

Like the author of Problems with crowdmatching + proposing an alternative, i’m not convinced that it is the right approach.

My background in short: I’m involved in many free software projects and try to find a way to sustainably fund public goods. I fully support your mission and share the vision, that FLO public goods can out-compete proprietary stuff with less funding. cooperation > competition

I think i generally understand how crowdmatching work, but i want a deeper understanding and get a feel for how it works in practice.

Problems i see:

  1. Funding of small projects is limited. They might have a dedicated community of enthusiasts that are willing to donate more, but they can’t. They would probably get more with Patreon.
  2. When a project is popular and has many patrons, many will not be able to afford to contribute. Only rich people can contribute. So it’s not democratic anymore.
  3. I would intuitively donate more to projects i care more about. Here i can’t decide how much to contribute. I might even not be able to afford contributing to my favorite project. I guess the theory is that my individual opinion is less important than the collective. It might protect me from stupid decisions to fund projects that will fail (like i did :grimacing:).
  4. matching has also the effect that people donate more (i guess). that mechanism is not used with crowdmatching
  5. When i contribute and that will bring a patron over their limit, my contribution is not matched. But that was my motivation to contribute. Do we have to communicate that when pledging?
  6. The theory is that crowdmatching will bring the project more patrons and funds than Liberapay or Patreon. A study with the first projects would be great.
  7. I like to say “when everyone contributes a little, we have a lot of funding”. consider every firefox user donates 0.10$. but they would have to donate a lot more…
  8. How would big projects work in practice? Consider these:
    • Wikipedia got 110,000,000$ in 2019 in donations, that equals 9,166,666$ per month. with crowdmaching, it needs 96,000 patrons which contribute $96 each. not sure if they want even more… they might need less with sustainable monthly donations
    • Firefox (400,000 people donated to Mozilla in 2019), Mozilla got about 16.000.000$ in 2018 in donations and about 430.000.000 in “Royalities” (is that the Google money?). total 450,000,000$, equals to 37,500,000$ per month. crowdmatching would need 195,000 patrons contributing $195 each (they would need less when they focus on developing firefox, so it’s probably overfunded and mismanaged. the TTS/STT project is nice, but can be split out)

Screenshot from 2020-08-26 00-03-57

Example projects:

Screenshot from 2020-08-26 00-48-38

These are all not failing projects, but they fail to get enough funding through Liberapay. But it would be worse in every single case with crowdmatching (with the same number of patrons).

Screenshot from 2020-08-26 01-13-53

Patreon don’t list projects publicly. The examples that they have, have 90% non-public income. The funding seems in all cases from enthusiasts. They market it even that way: “Patreon helps you build direct relationships with your most engaged fans.” Many creators are single persons, snowdrift seem to focus on projects.

3 Appreciations

Don’t have time for a full response but thank you very much for your consideration and feedback. I’ll going to cherry pick a few things to respond that I have immediate answers / time to respond to :slight_smile:

As it must be Free / Libre / Open source, and all the code is available: if a project grows beyond your means to stay contributing, the project is still 100% as useable to the contributer who can no longer, so what is the cost of dropping out? I would think that would be a good thing if it was getting so much support for development that one person had to drop out, as it would mean the project (that they can keep using) is just getting better support?

yeah I see your point here and conveying that might be important, but a big part of it all is to get sustainable funding for projects that we can all freely share and use. pledging at any point (even if you drop out) is a step in getting a project funded.

I would assume the goal is to help a project obtain the financial support it needs, and even after dropping out a person will have contributed to the project growing that support faster. Iinevitably some people will drop if the project grows enough, and thats a good thing (because it means better funding). Might be a bit sad for the person to get dropped but that also probably means that person doesn’t have the same finances as others who can stay contributing, which in theory is a good thing that someone who doesn’t have much surplus finances to keep using the (better funded) project without even contributing themselves.

The way it’s framed currently there wouldn’t be decision beyond ‘do you want to pledge or not’
The question ‘how much to pay’ on kickstarter and patreon are often quite arbitrarily set by the person running the campaign.

Absolutely! It’s quite theoretical until it’s actually put into practice

Yes, with crowdmatching at those numbers, each of those patrons would be paying less on average than are donating at their current rates. But there’s no extra incentive of matching for potential additional patrons. As you said: these projects are failing to get enough funding, so there needs to be more growth / incentives for more contributers (which in theory Snowdrift may help accelerate)

That’s what I got now and I’m sure other folks will chime in soon. Appreciate your thoughts and input @davidak

2 Appreciations

we miss out on that contribution.

i guess in classic fundraising naturally you have many contributors with small amounts and a few with big (think Wikipedia). with crowdmatching, you can have few small or many big. that’s the opposite!

that’s also the case with liberapay right now and they fail. the idea of crowdmatching is that your contribution get matched and that motivates people to contribute.

maybe not everyone would look so close like me if it is actually matched, but when someone does and is is not matched, they might feel scammed and want a refund! so we might loose 2 contributors and reputation every time this happens…

the options, yes. but i can still decide which option to choose or define a value freely.

See elementary OS:

They suggest that as a fair price, but you can pay what you want. Everyone contributes what they can. I think that is fair. You can even put 0 in the custom field and it is free. People that have enough money might not even think about how to get it for free.

1 Appreciation

(i can’t edit my original post)

I might believe in crowdmatching. Let’s make some assumptions!

Screenshot from 2020-08-27 11-18-58

Let’s assume $10.000.000 per month is enough for every project.

And assume that people are willing to donate $100 a month for a thriving project. (I donated 100€ per month to 2 universal basic income projects each, for over a year. I really believe in the idea. It would change our situation here completely!)

But will there be 100.000 patrons which are willing to contribute $100 a month?

(there would be still more people willing to contribute less than there are people willing to contribute $100. how to allow that?)

A good income might be $5,000 per month? So we can have 2000 employees. (A little less, because we have other expenses, but salary is the biggest. Some might need less salary.)

Does any single project need 2000 employees? Unity has “more than 2000”, but does Godot really need 2000 paid contributors to be the best game engine? Does Gimp need 2000 to be the best image editing program? Probably not?

Assume getting the project funded is more important than that everyone can afford to contribute.

You argued:

It is democratic, because everyone pays the same and people with more money can’t get more power over the project.

With a $100 contribution, only people with more money can afford to contribute and might have power over the project. They could threaten the project, that they collectively stop contributing when the project won’t act in their interest.


So, what i want is a practical calculation how crowdmatching would work for small, medium and big projects. Having that in the more detailed about pages would give people a deep understanding, that this is realistic.

We need a buffer, so the employed people can save money (or is that included in living wage?) and the project can also save for unsteady patron numbers. (Add X% to income and X% to total) A sustainable project should have savings for X month to continue.

Small project

An individual artist that creates web-comics
They need an income (living wage of $16 per hour = $2,770 per month) and have some expenses for material and webhosting ($100)
Needed budget ~$2870
With crowdmatching, 1700 patrons are needed, donating $1.7 each ($2890)

This is obviously David Revoy. We can ask him what budget an artist like him needs when they are 25 and single. (is that realistic?)

He has 900 patrons (852 on Patreon, 48 on Liberapay) and many other donation options.

Bigger project

Open Source Game Engine project
12 people employed with a good income ($5000 per month)
Infrastructure needed that cost ~$200.
Needed budget ~$60,200
With crowdmatching, 7,800 patrons are needed, donating $7.80 each ($60,840)

This is Godot. We can ask how much funding they would need to make the best game engine possible.

They have 1,583 patrons.

Medium project

Self-Hosted file hosting platform project
50 people employed with good income ($5000 per month)
Infrastructure needed that cost ~$500.
Needed budget ~$250,500
With crowdmatching, 15,830 patrons are needed, donating $15,83 each ($250,589)

This is Nextcloud. They have 41 employees, mostly in germany, probably with competitive income.

Big project

A linux distribution
500 people employed with good income ($5000 per month, $2,500,00 per year)
Other expenses of $1,000,000

Needed budget ~$3,500,000
With crowdmatching, 60,000 patrons are needed, donating $60 each ($3,600,000)

I’m thinking of Ubuntu. Canonical has 443 employees

.

Suse Linux has 1,750 employees, so $8,750,000 salary + say $3,000,000 expenses = $11,750,000

With crowdmatching, 110,000 patrons are needed, donating $110 each ($12,100,000)

So, again, does crowdmatching make sense?

I would think a linux distro should be funded by it’s users, but they can’t afford it!

3 Appreciations

This is a big (and important) conversation.

The core point is that the $1 per 1,000 is just a round starting point that gave us something to work with. Some of us are currently discussing a more flexible approach that we hope to present soon. But we’ve always had the idea that building flexibility in would be necessary post-beta (meaning after initially running live with the first couple outside projects).

I really appreciate you bringing up concrete numbers, something I’ve been resisting (for not all good reasons).

David Revoy is a good example and someone I’d love to get on board. But for initial starting point, we’re focusing on medium and large projects. Nextcloud is a good example, though there’s a challenge with projects that have a lot going on already because they are going to be somewhat conservative in trying out new ideas like crowdmatching.

Suffice to say, $1 per 1,000 isn’t “right” in some way, it’s just a simple starting point. And it will be best in dialogue with real projects to figure out the actual targets and potential. There’s a lot to learn. But that doesn’t affect the core concept of crowdmatching.

While FLO projects can probably get by with a fraction of the funding of proprietary, we can often sell them short. Even in the case of someone like David Revoy, he could probably use an assistant to help with all sorts of things he’d rather not have to do himself. True success (at the level of out-competing proprietary products) can require a lot of work in marketing, surveys, translation, and on and on.

If you didn’t review this yet, check out the old formula we initially ran as a fake-money prototype: https://wiki.snowdrift.coop/archives/communications/formula

Anyway, really soon @msiep plans to present his proposal for how to get crowdmatching to be more flexible, and it might address all your concerns. With one exception: no matter the other details, crowdmatching still inherently involves withholding potential donations until others join.

So, we’re asserting broadly that (see Liberapay etc) a small number of generous patrons is better than nothing but has no trajectory, no framing, little incentive, to grow to a higher magnitude. When I see a Liberapay project just doing something but seeming hopeless at competing with proprietary competition, I don’t feel encouraged to donate, I feel discouraged. And how do I set my level if I give? $2 to do something (but have no real impact)? $20 at some real cost to me and still no major impact? $200 at a big cost to myself and yet nobody else does this and nothing really changes? No, I want to be giving $20 as part of 20,000 people changing the world.

It doesn’t have to be those exact numbers, just the concept. I really do not feel capable of determining the “right” level to donate unilaterally in most cases.

To me, it all seems like a social negotiation. It’s the Snowdrift Dilemma. I’m sorta looking around at other people and thinking, “are you gonna shovel? I mean, I’m willing to, but but” and we’re all just standing around hesitating while a few folks are trying on their own and burning out. And joining them while nobody else makes the change isn’t that compelling.

1 Appreciation

maybe projects that have no real funding? so they don’t loose money when crowdmatching needs some weeks to get the needed contributors to have an impact

i contribute to NixOS a lot. it’s the most advanced linux distribution and has the most up to date packages. in the last month, we had 415 contributors, but not a single paid employee. we struggle to review PRs! we are very innovative, so we might try it. we had no discussion about paying someone yet

i’m also involved in apertus°. it’s mostly 2 very dedicated people who build a digital cinema camera in their free time over the last 10 years. having 1, 2, 5 paid employees would make a huge difference!

also, you mentioned inkscape. are there concrete discussions going on? would be a good fit

i’m very curious and would love to see a working solution

you said that it is a feature and i agree

i have thrown more than 300€ into liberapay and it probably has no impact since it can’t pay a single salary for any project. maybe cover some expenses. i’m not rich, i can’t fund all FLO alone

yes. i actually have burnout and had to quit my job, was in rehab and work now part time to have more time to contribute to free software. i see all the problems in the world and want to change something. but i can’t do it alone! sometimes i think i’m the only stupid person who does something and everyone else don’t cares (but i don’t give up)

1 Appreciation

The big concern I have is that seeing people like you put in a lot and have burnout and see little change in the status quo… this is majorly discouraging for others considering picking up a shovel.

Your time and money did something for sure. In the case of Snowdrift.coop, I’ve spent thousands of dollars and thousands of unpaid hours, and if donations mean that I still don’t get paid but at least don’t have to spend more money myself, that’s something.

Most FLO projects are at the minimizing-losses stage. It’s real value to minimize losses and to have an expense budget. And even just part-time pay would allow me to worry less about my music teaching business and be less likely to burn-out over financial concerns. So, again, while I completely support having milestone goals, I really don’t think funding in most cases is all-or-nothing in practice.

The shift with crowdmatching overall is toward folks like you and I ending so much self-sacrifice and burning out. We put put in a little toward projects that don’t do that much but at least exist. We might put in a lot as part of larger crowds that really change the world. Collective action!

1 Appreciation

thanks. some random thoughts:

when you consider allowing doubling the pledge, so people with more money can contribute more, why not also allow half pledge, so people with not so much money can still contribute and get matched, even when one share is quiet high? is this the solution? :astonished:

for communication, you might not have to explain the whole concept in all detail on the main page, but only show “contribute NOW to get matched 1.4x”. many people might be fine with “they invented a complex formula that allows getting matched by the crowd while being flexible how much you contribute”. they might look at the details page, but think soon “this is too complicated for me, i’m not good with math” and that might be fine

for better understanding of the mechanic, i would like to see an interactive visualisation where you can change all the parameters. it could be on the details page

i had the idea, that people who drop out form a new pool that start low again, until they drop out there again… so you might have 3 pools, one high, middle and low, but it don’t stop there… you will have many pools and matching does not happen. this is a stupid idea

When we first made that prototype with that ability to have multiple “shares”, we had the starting point at a “mil” as in a hundredth of a cent. So, there was no need to have partial pledges of that. But the whole extra formula was too complex to explain, and it had the problem of people not having a basis for considering what the “right” pledge level was (testers were putting in 100-fold pledges and such). Anyway, it’s not the new proposal I’ve been hinting at.

Otherwise Snowdrift Wiki - Limits to Pledge Amounts describes the other ideas of ways to allow patrons at lower levels (like less-frequent participation and breaking up large projects into sub-projects). But that may not be needed if we adjust to a different flexible core model.

This shows some good thinking about the puzzles we’re facing. We say “crowds” though generally :wink:

1 Appreciation

On the new more-flexible goal-based proposal: We will have further discussion, and then if we have consensus, we’ll make plans to update the wiki, the main site presentation, the UX, and a blog post etc.

For now, I captured our meeting notes about the proposal from @msiep on the wiki: https://wiki.snowdrift.coop/resources/meetings/2020/2020-08-28-crowdmatching-update-proposal

UPDATE: I edited that page to capture more from my memory of the meeting and add the illustration from the screen-share. (Also fixed an initial typo of ‘crowdmathing’ that I had made).

I really think that this direction is super promising. The next steps will be to map out the options starting at the simplest version. We could then decide which aspects need to be determined for MVP. We should also document the rest of the possibilities as ideas we have available to us as we gather real-world operating metrics and feedback and approach the conditions under which various options would come into play.

1 Appreciation

Great post, @davidak. You’ve convinced me, once and for all, that a one-size-fits-all pledge amount simply won’t cut it.

Personally, I’ve never really imagined Snowdrift working that way, and I didn’t realize that was the original intention - I thought you’d always be able to pledge multiple times. In fact, I also had the same “partial pledge” idea, so you can be, say, half a pledge, and others only see the number of total pledges.

Anyway, back to your original post.

A minor nitpick:

Huh? Sure it is. That’s the point. You even say so in the next line:

So matching motivated you to contribute, rather than not contribute. That’s donating more. More than you would have.

1 Appreciation

About the process needing to be “democratic”. Everyone pledging the same amount, could certainly be called democratic, but I’m not sure that a variable-pledge system is ever strictly democratic. However it doesn’t need to be full-on capitalism either. Wealth isn’t always a bad thing, it does often arise from behaviors that do deserve to be rewarded (patience, hard work, high-demand work, financial smarts, providing something to the world) - just not as often as we’d like. Wealth also often arises from things we’d rather not incentivize (greed, birth lottery, illicit activities, etc.) which is why the rich get a bad rap.

But you hit the nail on the head here:

That’s the additional factor in how much one donates. If we could filter out wealth, and voting-with-dollars was only based on how much you care about something, wouldn’t that be better than democracy? I feel like democracy is a compromise we accept as a kid, when we realize that we can’t figure out a system where “but I want it more!” is what matters.

So, since it’s come up here before, I love the idea of Quadratic Voting as an actually-workable solution to this. Paying more money means you want it more (to the point of sacrifice), which we want to base your power on, but it can also mean other unrelated things (you’re wealthy) which we don’t want to base your power on. So just when it seems like democracy (everyone gets the same power) is the only solution, Quadratic Voting comes in to say “How about we let additional money mean additional power… but only by a little bit? Like, enough to feel “worth it” for the rich to give more, but not enough to let the rich take over. Mostly democracy, with a little non-toxic amount of capitalism sprinkled in”.

If we could work that into the system, that’d be awesome.

3 Appreciations

I think this is an understatement.

I’d say that rather than a few big donors, there’s a TON of big donors, a LOT of HUGE donors, and BOATLOAD of “big” ones.

In fact, I suspect that we should start not with the assumption that the donor list is a flat-ish line across the spectrum of wealth, but instead assume the contributions follow the bell curve. This means that to achieve the successes you reference, allowing variable donations matters a lot more than we think.

Wikimedia

To test my hypothesis, I had a look at the list of big donors toward that Wikipedia total you gave.


(these numbers are from 2018, but in 2019 they only go up)

Wow! Here’s the list of big donors, some of which are single individuals, contributing no less than what would be like an $83 per month Snowdrift pledge. Look how long it is!

A bell curve has a lot going on on the outer edges, and these look like outer-edge numbers to me. A thin extreme - 21 at the far-right tip, 21 further in, then shoots up to 185, then to 1,529 donors.

A bell curve has a massive lump in the middle, which is why despite the huge edge numbers, that “mean” line can be a rather unassuming amount. Sure enough, in Wikimedia’s case that lump is an area of 7 million people, bringing the mean to the $15 mark. A pretty modest mean - but if everyone gave that, we’d be $14 million dollars short!

Sounds like a bell curve to me.

Alas, we can’t know for sure since the exact donation data is confidential.

1 Appreciation

matching has also the effect that people donate more (i guess). that mechanism is not used with crowdmatching when everyone gives the same

But that is not a limitation anymore with the new proposal.

Hmm. Everyone definitely gives more (than they would have) with the original crowdmatching, because the system makes them. So I guess what you’re referring to, is that it doesn’t encourage them to manually give more simply due to the psychological effect of someone offering to match the donation, the way it would with standard fundraising? (But that’s only because we’re making that behavior built-in, so it’s still there, just not optional.)

exactly

the overall funding is not more when you pledge and someone else drops out.

but let’s not discuss the old proposal.

To be clear, nobody suggested that $1 per 1,000 patrons would be a fixed number for all forever. But we were previously only at this place (stated as the pre-newer-proposal view I might have said):

“The old logarithmic shares-matching formula has good aspects but is too confusing to explain and isn’t likely to be returned to (though we did discuss ideas like after initial pledging we could have an extra click to double-pledge and such set up to offer multi-level pledges but with constraints). We know that projects of different sizes and different level of crowd enthusiasm will need ways to set different match-rates or something, but we’re holding off on adding too many options. We’re going to launch with simple in-or-out with the round starting number and iterate based on real-world feedback.”

In short: we did not have consensus about how/when/whether to add what sort of flexibility. I think the new proposal sets us on track to achieve consensus, but we haven’t nailed down the precise plan (which elements will be MVP, which will be basically settled, which will be documented as options but wait-and-see before choosing direction).

Alright, I finally found some time to chime in here. [Un]Fortunately, other people have already said many things that I would have — I’ve left those out of this post, I hope.


I think the theory is sound, but we can’t really know without trying it… in some sense, answering this question (hopefully with “Yes!”) is the main goal of the project.

We need a buffer: “A sustainable project should have savings for X month to continue.”

This also ties in with practical implementation concerns: while using credit cards, we charge in arrears — postponing the actual charge until fees are acceptably low (we chose 10%, which works out to a $3.79 minimum charge).

Take Snowdrift.coop itself: By my back-of-the-napkin math, we have ~$300 in pending donations and will have ~$350 by the time we actually run a payout. We needed ~4 years of buffer… but of course, we had other projects to pledge to, some patrons would reach the minimum faster than that (or if we’d done any outreach/publicity).

matching has also the effect that people donate more

When I’ve seen these campaigns, the marketing usually goes, “${Rich person} has agreed to match the first ${amount} of donations (until ${date})!” But what if that amount isn’t reached? Then you have the same situation as early crowdmatching — withholding a donation until others join in.

@msiep I’m not sure whether the result would be any good, but it would be interesting to try and frame your approach in these type of words (the “usual marketing” ones).

[Small projects] would probably get more with Patreon.

Perhaps, but donations at that scale won’t achieve the vision of FLO out-competing proprietary stuff anyway. It’s okay if Snowdrift does not serve every project perfectly; we can exist alongside feel-good donation sites (where the primary goal is to express appreciation, not pay salaries).

Crowdmatching always protects you from “useless donations”, by definition; if you’re not withholding down to (some) level proportional to the crowd size/level, you’re not crowdmatching.

The design where crowdmatching levels are fixed, and you just choose which projects to support + a budget for the site overall has the additional nice property that you can pledge to as many projects as you want without raising your limit. This means you can pledge to all the projects you’d like to support, and let the system direct your money to the projects with more other supporters, where it will make more of a difference. But this comes with some trade-offs, and it isn’t necessary for Crowdmatching.

It doesn’t have to be this way. I’ve been experimenting with some ideas for a multiple-tiered Crowdmatching formula where the more money a project receives, the less each individual patron gives (overall — the actual curve has both upward and downward slopes).

I’ll write a separate post about these later; they are not worth talking about until we’ve discussed @msiep’s idea, which they build on.

First, I am not sure that Wikipedia is particularly a representative project. I think it’s something of an ideal-case FLO public good, where it’s so widely used as to be a household name (and, in my experience, more well known than the concept of wikis themselves — I’ve seen people reference stuff they read “on wiki”, not realizing that -pedia is not the only one!). So, its ability to solicit donations may be similarly abnormal.

I will also note that we do have a resident expert on Wikipedia, although I don’t know how much he’s at liberty to share.

3 Appreciations

When the amount is not reached, that what is reached is still matched, as i understand it. Why would i hold back. I will donate because my donation is doubled.

(GitHub Sponsors has matching for 1 year for any project or person. But the amount is not limited. It motivated me to donate there. But it’s also just a small crowd of 5 people in many cases.)

Maybe the limited amount and pledge time also motivates people through FOMO. But that is kind of a dark pattern. So it could be questioned if crowdmatching works as well as limited matched donations.

Could we get some scientists involved to do an accurate study? There is the Max Planck Institute for Research on Collective Goods in germany for example.

Yes, i think we can have a call to action (CTA) on the project page, like “Donate now to get matched by the crowd!”, so people actually consider it. But we don’t want them to feel anxious.

No, it’s not representative. Wikipedia and Firefox are the only huge FLO projects i came up with and they are quiet different.

But Wikipedia is probably still funded by a fraction of it’s users and crowdmatching can help to get either more money or only more donators that each donate less. and they don’t need that annoying campaign every year that probably also costs resources

Yes, “Wiki” is just short for Wikipedia in that case. They don’t know that a Wiki is a stand-alone concept.

Not you — it’s the rich person who’s holding back. If they say they will match (1:1) the first $10k, and people only donate $4k total, then the rich person withholds $6k.

1 Appreciation