Project Wonderful as an economy

If you've been following my adventures trying to advertise my Web comic, you'll have heard about my misadventures with Google, and more recently the collapse of my Project Wonderful ad box bids.  My thinking is that this may be a structural problem with Project Wonderful - it's set up in such a way that when someone like me starts lowering bids, that causes other people's income to drop and then they lower bids too and the whole thing can spiral out of control, leading to ridiculously low bids and a non-awesome situation all around.  One thing that would help would be if the highest bidder paid their actual bid instead of "just enough to beat the second-highest bidder"; that way half as many people (i.e.  only the top bidder instead of the top two bidders) would be in a position to affect the amount of income seen by the seller, and so there'd be a lower multiplication factor and more chances to end the chain reaction.

"Highest bidder pays own bid" would also eliminate the current problem that the "just enough to beat" algorithm isn't even properly implemented in the current system - at present, you can reduce your payment while still beating the other bidders if you arrange things so that your bid hits maximum, and as a result, you have an incentive to lower your bids more often to be sure that bids do hit maximum.  I've been in touch with Ryan from PW about that one and he says he'll fix it - eventually - but it sounds like a difficult, complicated problem with a lot of side effects, so it may be a while yet before it gets really fixed.  I think that would also tend to reduce the chain reaction problem.  The basic issue is that there are a lot of feedback loops that cause small fluctuations to be magnified.

I've been trying to come up with solutions for the feedback loops based on considering PW as an economy like other economies.  What would Canada be like if the government decided to abolish income tax, and almost all other taxes...  but also to set the GST at 25% and eliminate the current Byzantine system of GST exceptions?  That situation is kind of what Project Wonderful is like at the moment.

I figure that there are a few big-time sites in PW and a whole lot of small-time sites; my site is one of the small-time sites.  As a small-timer, I can afford to put in maybe $0.50 per day of real money - the exact amount isn't important, but it's on the order of less than a dollar.  Right now it's zero, but the average over time is in the $0.50 range.  I spend the money advertising, on other small-time sites like my own because that's much cheaper than advertising on big-time sites.  Other small-time sites like my own advertise on my site, for the same reasons.  Big-time sites do not advertise on small-time sites because, in general, they don't need to advertise at all.  The amount of money I spend on PW advertising is the $0.50 I put in per day, plus whatever I can make from other people advertising on my site, which in turn is 75% of what other people spend advertising on my site.  Ryan, the PW operator, takes a 25% cut.

In general, small-time sites do not take out a profit.  I certainly don't, and it seems clear that I can't reasonably expect to ever be able to take out a profit as long as I'm a small-time site.  That's actually an important issue, for tax reasons:  if anyone asks, I can solidly document that no reasonable person could expect my site to run a profit, and for that reason, it's not a business, and I don't have to pay income tax on it.  If I didn't spend the money on advertising, and tried to take it out of the system instead, my own traffic would fall enough that nobody would want to advertise on my site and I wouldn't have any money coming in to take out.  So it's not real money from my point of view - it's not a profit I can spend on other things, and once I click the Paypal link to put money into my PW account, that money is dead to me.

Assuming other small-time sites operate the same way I do, what does that mean for the fate of money in small-time sites' PW accounts?  It means, first of all, that all of that money eventually ends up in Ryan's pocket, because he's the only one for whom it makes any sense to take money out of the system.  Furthermore, it means that the actual amount of his cut doesn't matter.  At present it's 25%.  Every dollar I put in will eventually result in four dollars' worth of economic activity; the money gets spent an average of four times before it goes into Ryan's pocket.  But the game would be played the same way, from my point of view, for any nonzero cut!  If Ryan's cut were 10%, then each dollar would be spent ten times; but I'd still be putting in about $0.50 per day because that's what I can afford, and I'd still be buying about as many incoming clicks as I can sell outgoing clicks because on average, everybody is doing that for the system to balance, and so on.  Similarly if Ryan's cut were 90%.

There is one thing that changes for small-time sites depending on the cut, though:  the average nominal bid level changes.  If Ryan's cut is 25%, and I can afford to put in $0.50 per day, and we assume everyone is the same as me, then it's not hard to do the math and see that my $0.50 must balance with the 25% taken out of my incoming money stream in order for the total to match my outgoing money stream, and so my outgoing money stream - the total amount I spend bidding on other people's sites, which is equal to the total amount other people spend bidding on my site, must be $2.00.

With a 10% cut, my revenue stream is $5.00 - that's $0.50 from me and $4.50 from others.  With a 90% cut, it's $0.55 - that's $0.50 from me and $0.05 from others.  Understand that I'm getting the same amount of advertising for the same price of $0.50 per day in all cases - those are external constants that aren't going to change.  The number of clicks I buy must be equal (on average) to the number I can sell, and I'll put in as much real money to pay for those as I can afford, and it all averages out.  The amount of real money I take out is zero in any case.  The amount of real money Ryan takes out of the system is fixed, too - on average, it must be exactly the amount everyone puts in.  But the nominal funny-money price changes depending on his percentage cut, which determines how many times the money cycles through the system before he takes it out.

If you're of an economic bent, think about the "reserve requirement" in the banking system.  When banks take deposits, they are allowed to loan those deposits out again except for a percentage called the "reserve".  The government tells the banks what reserve percentage they must use, and that determines the money supply and is part of what determines the velocity of money - how fast economic activity occurs.  With a reserve of say 25%, money gets re-loaned four times.  With a reserve of 10%, it gets re-loaned ten times.  The actual reserve requirement in the USA at the moment is 10% for chequing accounts and 0% for everything else.  It's zero in Canada (but that's the mandated minimum; the banks actually do keep nonzero reserves), 2% in the Eurozone, and you can look up lots of other numbers in the relevant Wikipedia article.  Governments in general will lower the reserve requirement when they want to stimulate the economy and cause there to be a lot of economic activity; they will raise it when they want to slow things down.  Low reserve requirements are inflationary and high reserve requirements are deflationary.  Project Wonderful has something that behaves like a very high reserve requirement, and guess what, Project Wonderful's economy is deflated right into the toilet.

But Project Wonderful isn't composed entirely of small-time operators.  There are also the big sites, the Questionable Contents and the Girl Geniuses and so on.  That's where the percentage cut becomes significant, because the big-time sites do not operate like mine.  The way they work is that there are big-time buyers, like Dirty Microbe and Tiny Warz.  They do NOT sell ads of their own - or at least, that isn't a significant part of their business model.  They're actually selling something else to make money, they have real money to spend on advertising it (dollars per day per ad space; upwards of ten cents per click, the same as they'd be paying Google) and they spend it on the sites of big-time sellers.  As I write this, Jeph at QC is getting bids of $8.30 for each of two square slots.  I don't have access to Jeph's or Ryan's account books, but it seems pretty clear that that money isn't staying in the system.  If it were, Questionable Content ad buttons would be outbidding mine all over the place.  Instead, it looks to me like what's happening is that Dirty Microbe ponies up their $8.30 - of real money they deposited through PayPal, not funny money someone else deposited - and Ryan takes out his $2.075, and Jeph takes out his $6.225 as real money, and that's the end of that story.

So what happens if Ryan changes his cut there?  Then it makes a real difference.  Ryan doesn't really get 25% of money spent - he gets all of money spent on sites like mine (because that money just gets re-circulated) and 25% of money spent on sites like Questionable Content.  Changing the cut percentage would change the take from the big-time sites, without directly changing the take from the small-time sites.

It would have an indirect effect, though, because of the fuzzy boundary between big-time and small-time sites.  If the cut were to decrease, nominal bids on small-time sites would increase because they're paid for mostly with ad revenue.  Nominal bids on big-time sites wouldn't, because those are paid for with real money coming from outside the system and leaving the system after one pass.  Changing Ryan's cut changes his revenue from big-time sites without changing their nominal bids; it changes the nominal bids on small-time sites without changing his revenue from them; and it changes the interaction between big-time and small-time sites.  At the moment, small-time players can't reasonably afford to bid on big-time sites at all, and big-time buyers consider small-time sites to be beneath notice.  If the nominal bids on small-time sites could be raised without changing anything else, then they could start interacting with big-time sites.  And, what do you know, changing the nominal bids is exactly what happens when you change the 25% cut.

But merely changing the cut is a problem because it'll eat into Ryan's profit from the big-time sites, which I suspect is his main source of profit.  Every $2.075 he pockets from one slot-day on Questionable Content covers an awful lot of slot-days like the ones on my site, for which he gets at little as $0.005 (25% of the current bid price $0.02 per day).  In fact, I suspect he makes a loss from sites priced like mine after the cost of providing the Web services is covered.  There are a lot of cheap slots in the system but they pay very little.  The revenue from the big-time sites is directly proportional to the cut, at least as long as the cut doesn't get so very big that it would change the big-time sellers' behaviour.  I'm sure Ryan can't take much less than 25% without making his business unprofitable, or he would.  So, what to do?

I think it might be a good idea if Ryan lowered his cut on transactions to much less than its current level - possibly 5%, or maybe even 0% - but if he charged the current 25% or maybe even a little more, like 30%, as a fee on all withdrawals from the system.  What would happen then?

Think about the analogy of the Canadian economy - it would be like instead of having a 25% GST as the only tax, having a relatively small GST and a significant income tax.  Which, in fact, is the way things really are.

This scheme would provide a strong incentive for the big-time sites to NOT take their money out, but instead spend it within the system, advertising on smaller sites.  For your dollar of ad revenue, you can get $0.75 of real money, or $1.00 of advertising; so if advertising is something you have any interest in buying, it looks like a good deal.  It would mean that those of us who bid with money we got by selling ads, would be at an advantage relative to those of us who bid with money we deposited through PayPal.  I don't think that anyone would really be depositing more or less money, at least not among the small-time buyers.  I don't change my ad budget depending on price, I decide how much I can spend and then try to get as much advertising for that amount as possible, and I expect others do the same.  But I think that probably more of that fixed rate of money being deposited (and certainly no less) would be captured and re-used (and eventually end up in Ryan's pocket) before it left the system.  Nominal bids on small-time sites would rise, bringing them into line with the nominal bids on big-time sites, and that would tend to erase the distinction between small-time and big-time sites.  We might see more consistent prices per click and per impression between large and small sites, whereas at the moment, small sites are MUCH cheaper on both measures despite the expectation that the free market should level that out.

There's also the important psychological factor of bid-price sticker-shock - it's not at all fun for me to log into the system and see prices of $0.02 on my slots, never mind that I have no intention of actually cashing out that money, and the low prices on my slots correspond equally to the bargains I can get on other people's slots.  Even having analysed the situation as carefully as I have, and knowing that they're just artifacts of the high cut rate, the low nominal prices are decidedly not awesome.  Higher nominal prices would be closer to awesome.

Current system:  Ryan gets all the money from small-time sites and 25% of the money from big-time sites.  With a 5% cut on transactions and 25% on withdrawals:  he gets all the money from small-time sites and 28.75% of the money (that's a 5% cut followed by a 25% cut) from big-time sites.  With 0% on transactions and 30% on withdrawals:  he gets no money from small-time sites (but I don't think they were ever a big deal anyway) and 30% of the money from big-time sites.  Only Ryan has the data to know whether those options would make business sense for him, and nobody can really know without trying it exactly how the buyers and sellers would react, but it sounds to me like drastically reducing the cut on transactions and replacing it with a cut on withdrawals would result in a system that would be more awesome than the current system.

Why post this here instead of emailing it to Ryan?  Mostly because I don't think it's at all likely he'd really use it, and I think it's probably more interesting to my readers than to him.  Being an armchair entrepreneur is fun.