# Can inflation occur in a positive-sum game currency system such as the Stack Exchange reputation system?

As most active users of this site know, the way the reputation system works is that everyone starts out with 1 reputation and a user can earn more reputation by asking good questions and writing exceptional answers (as well as other methods), and a user can lose reputation by writing bad questions and answers. This way, users can make a profit or take a loss without taking/giving currency from/to another user. Can inflation happen with a system like this? As more reputation is gained, does it mean less?

• It's not really a currency because it's not really a medium of exchange; it occasionally gets used for bounties but that's about it. Apr 15 '19 at 10:17
• The reputation levels for several privileges are different across StackExchange sites. Aren't those changes related on how many people are on different sites and how much reputation they have? Haven't those levels changed with time as sites developed? All answers here are assuming that those levels (prices) are fixed, but I'm not sure that they are.
– Pere
Apr 15 '19 at 10:29
• @Pere indeed, new sites have lower thresholds for several moderation privileges, simply because there are too few people that could otherwise use them.
Apr 15 '19 at 11:47
• Therefore, there is some kind of inflation on price of privileges, although its behavior can be very different from inflation in "real" economies.
– Pere
Apr 15 '19 at 12:01
• Besides the minor tangible uses (bounties), reputation points also have value in showing reputation, and that might be subject to inflation over time. For example, if a recruiter saw that a job candidate has 1000 reputation in 2013, he'd value that more highly than if he did in 2019. Apr 16 '19 at 9:15

In the Stack Exchange reputation "economy", we can think of the reputation points as "money". And the "goods" are:

• Downvotes — cost 1 point each.
• Various privileges — zero cost, but unlocked only when the user reaches various thresholds.

To my knowledge, these prices (1-point cost of each downvote and privilege thresholds) have never changed. And so, there has never been inflation. What has merely happened is that "wealth" (the total stock of reputation points) has kept increasing.

Response to Pere, who writes:

Newer sites have lower thresholds for several privileges - for practical reasons - and those thresholds rise as the site expands. For example, being able to create tags "costs" 150 reputation in Economics SE, 300 reputation in Cross Validated SE and 1500 reputation in StackOverflow. Therefore, privilege prices are subject to inflation as sites grow.

Each individual SE site is its own individual "economy" that is sealed off from all other SE sites.¹

And so, whether "prices" differ across SE sites does not matter. What matters is whether "prices" have changed over time, within each SE site.

To my knowledge, the privilege threshold "prices" on Economics SE have never changed.

And it also seems that for the original, flagship site (Stack Overflow), "prices" have also not changed for many years. We can, for example, compare today's "prices" with those from 2013:

As you can see, all prices have remain unchanged since 2013. There has therefore been zero inflation since (at least) 2013. The only change is the introduction of a "new good", namely "Access to site analytics", which is priced at 25,000 points.

Response to @kbelder, who writes:

What about bounties? I bet those have historically trended upwards in value

The "prices" of bounties have likewise been fixed. Bounties may be offered at "prices" of 50, 100, 150, ..., and up to 500 points. I believe these prices have been fixed since the inception of Economics SE and at least since 2013 for all SE sites.

Moreover, bounties are pure transfers from one user to another.² It may be true that thanks to increased overall "wealth", the average value of a bounty has risen (i.e. in the past, most bounties were offered at close to the minimum price of 50; whereas today higher prices are offered). However, we do not consider this "inflation". Instead, we consider this to be real wealth effects.

Analogy: Say that in 1919, the New Yorker on the street might on average give \$0.20 (in 2019 dollars) to a homeless person. In 2019, this might instead be \$1 (in 2019 dollars). The average real transfer from a New Yorker to a homeless person has risen five-fold in real terms. This is not inflation. Instead, it is a consequence of increased overall wealth.

Footnotes.

1. To my knowledge, there is one thing that links reputation points across the various SE sites. This is the automatic +100 points for established user accounts that are "trusted" on the SE network. I'm not sure how this works. But I would say that these +100 points are relatively inconsequential and may be safely ignored. They would perhaps be the real-world equivalent of an immigrations officer giving a "trusted" visitor \\$1 the first time that visitor arrived in that country.

2. There are actually three things that can happen to a bounty. First, the full bounty is awarded (I believe this is the most common case). Second, half the bounty is awarded to the highest score (provided that answer has a score of at least +2 — see Economics.SE bounty rules here). Third, no bounty is awarded. In the latter two cases, we do not quite have a pure transfer; instead, we have a leaky bucket where either half or all of the points are "lost to society".

• @Jodast: I think what you are alluding to the idea of "consumption externalities", where for example if everyone else has a car, then my car is somehow less valuable. To the extent that these effects exist, then yes, you may be right. However, I believe in discussions of inflation we usually leave out these effects.
– user18
Apr 15 '19 at 2:34
• And so, suppose the prices of everything stay fixed over the course of decades while everyone gets richer (and can afford more stuff). Then one could argue that there has been inflation, because though prices have stayed fixed, the value of the same stuff has actually gone down because now everyone can buy lots of them. However, economists who study prices and inflation will usually simply say that there has been no inflation.
– user18
Apr 15 '19 at 2:34
• "buying lots of something with more money than before defines inflation" — I think here you are making a separate point of confusion. In particular, you are confusing correlation with definition. We define inflation as prices going up. In the real world, we observe the following correlation: When people buy more stuff, then (as a result of various forces) it tends be that we also have inflation. However, people buying more stuff is not the definition of inflation.
– user18
Apr 15 '19 at 2:50
• We can see a bit of this by looking at the Econ 101 supply-demand diagram. Usually, when demand increases, the (equilibrium) price also increases (this is the commonly-observed real-world correlation just mentioned). However, here, the supply of downvotes is infinite. It is horizontal and fixed at a price of 1 (rep point). So, even when demand changes, price does not change.
– user18
Apr 15 '19 at 2:50
• What you're talking about is price Inflation - this is different from other definitions of inflation, such an an increase in the money supply or in the velocity of money in the economy. Price inflation is usually just symptomatic of monetary inflation. In this case there has been huge monetary inflation since the total number of reputation points in existence continues to increase, as does the required work to acquire them. If anything there has been price deflation in that you need to do much less work now than before to farm rep. Prestige value per rep has also deflated, likewise.
– J...
Apr 15 '19 at 16:16

No, there is no inflation.

Inflation is caused by a relative increase in the amount of currency, relative to the amount of scarce goods and services available.

As you note, there is a continuous increase in the amount of "currency" here - reputation.

However, reputation is not spent on scarce goods and services. There is no meaningful limit on the number of people who can attain any given rep-based privilege. So the rep-level for a given privilege does not need to increase, when the number of people with high rep increases.

The rep-level barrier for privileges is to ensure that on the whole, the people who attain a specific privilege are familiar enough with the workings of the site, and have shown enough good behaviour, to wield it responsibly. This is not affected by the number of other people who have also reached that specific rep level.

### Yes, there is inflation.

StackExchange was created in 2008. Anybody who had 5,000 (arbitrary value) reputation points in 2009 would have been perceived as an outstanding member. Now, having accumulated 5,000 points in ten years of participation on a much expanded site looks less impressive. Thus, achieved respect per reputation point has decreased.

The respect is not related to the absolute number of reputation points but to belonging to some top percentile of active members. As the necessary threshold increases, the value of an individual reputation point diminishes. I would call this inflation.

### P.S.:

KennyLJ has commented that he conscientiously ignored consumption externalities and positional goods because they are tricky and generally ignored in Economics. If I interpret his comment correctly, these aspects should not be considered in the context of questions about inflation like this one.

However, I think that this "consumption externality" is the key element for understanding StackExchange. It's not an externality, but the product itself. Badges and privileges are secondary.

• 5,000 rep can still be pretty impressive on a site in beta. For example, SE.Quantum computing just celebrated its first year, with the post specifically listing the users who had 2,000+ rep (of which there were only 8).
– Nat
Apr 15 '19 at 19:51
• @nat it's possible then that inflation is relative to the site being used Apr 15 '19 at 23:46
• I have already addressed this point in the comments to my answer above. To the extent that we allow for "consumption externalities"/"positional goods"/etc., then yes, one may argue that there has been inflation. But to the extent that we think only of the actual "goods" that may be purchased using reputation points (these "goods" being downvotes and the various privilege thresholds), then there has not been any inflation.
– user18
Apr 16 '19 at 1:04
• "Consumption externalities"/"positional goods"/etc. are a tricky thing that are — rightly or wrong — ignored in conventional economic analyses of real-world inflation. I have therefore chosen to do the same in my answer above. Whether and how "consumption externalities"/"positional goods"/etc. should be incorporated in economic analyses of inflation is a good and tricky question, but one that I think should be asked separately.
– user18
Apr 16 '19 at 1:04

Yes, there is some kind of inflation on Stack Exchange.

As pointed in other answers and in comments, reputation can be seen as a currency to buy privileges. Newer sites have lower thresholds for several privileges - for practical reasons - and those thresholds rise as the site expands. For example, being able to create tags "costs" 150 reputation in Economics SE, 300 reputation in Cross Validated SE and 1500 reputation in StackOverflow. Therefore, privilege prices are subject to inflation as sites grow.

However, it should be noted that the economy of privileges is not a free marked one but a planned one, because privilege prices are not related to demand but set by a central authority in order to allocate the desired number of privileges.

Interestingly, bounties economy is indeed a free market one, and it would be interesting to study the evolution of the bounties offered.

• It's not a purchase because you still have the reputation after you gain the privilege - there's no exchange involved, Stack Exchange gives it to you by virtue of you having it. Also, you can gain reputation far in excess of the amount of privileges you could possibly acquire. In a "normal" economy, you could (in principle) continue to purchase goods as you acquire more money - for example, I just purchased several thousand dollars of stock this morning, but the fact that I purchased stock today doesn't mean that I've gotten closer to some limit of goods I can acquire. Apr 15 '19 at 16:09
– user18
Apr 16 '19 at 1:11
• @KennyLJ - In your answer you claim that prices have not changed from 2013, which I assume to be right, but you can't use that to claim that prices have never changed. As far as I can see, sites in beta state have some differences compared to normal sites and those sites have some differences with StackOverflow. I assume that some changes happen as sites grow - maybe automatically when they leave beta state.
– Pere
Apr 16 '19 at 8:06

tl;dr- Sure, we can say that rep undergoes inflation/deflation. However, like the US dollar or any other currency, we'd need to qualify what market basket we're comparing it to, as claims about inflation/deflation are meaningless without a reference. In the case of StackExchange rep, the relevant market basket is less obvious, so it'd need to be qualified for a claim about inflation to make sense.

## StackExchange rep can be said to undergo inflation if the meaning of the claim is established.

Even US currency doesn't experience inflation ideally.

For example, this website claims that $$1\,\text{USD}_{1950}$$ (i.e., $$\1$$ from 1950) would be $$10.55\,\text{USD}_{2019} .$$ Presumably that means that $$\1$$ back in 1950 could've bought about $$\10.55$$ worth of similar products today.

Except that isn't true for technological products. $$\1$$ will get you far cooler stuff today than it would've in 1950! If we look at the cost of floating-point operations per second (FLOP/s) over time, it'd seem that the US dollar has undergone wild deflation.

The trick behind discussing "inflation" seems to be to put it up against a reasonable market basket. We can then say that inflation between two times $$t_0$$ and $$t_1$$ would be$$\left[\text{inflation factor}\right]_{t_0 \to t_1} ~ \equiv ~ \frac{ \sum_{i}{w_i \, P_{i}\left(t_1\right)} }{ \sum_{i}{w_i \, P_{i}\left(t_0\right)} } \,,$$ where

• $$w_i$$ is the weight of market-basket item $$i ;$$

• $$P_i\left(t\right)$$ is the price of market-basket item $$i$$ at time $$t .$$

I assume that this website claimed an inflation of $$\frac{10.55\,\text{USD}_{2019}}{\phantom{0}1.00\,\text{USD}_{1950}}$$ because it was looking more at stuff like food than FLOP/s.

Say a newscaster gets on TV and says that the US dollar is undergoing wild deflation, implicitly referring to a basket composed of stuff like FLOP/s. They'd probably get a lot of flack for false reporting, right? People would find it deceptive to speak of inflation in that sense since it's not what's normally meant by it.

Likewise, we can say that reputation on StackExchange has a changing value over time. Presumably a bounty for $$+500\,\text{rep}$$ would have less power to entice @JonSkeet now than it may've at StackOverflow's inception.

Then if we judge inflation of StackOverflow rep by its power to entice @JonSkeet to deliver some service, e.g. answer a question about how to write a for-loop in C#, then that inflation would be$$\left[\text{inflation factor}\right]_{\text{2009} \to 2019} ~ = ~ \frac{ P_{\text{answer}~\texttt{for}\text{-loop question}}\left(2019\right) }{ P_{\text{answer}~\texttt{for}\text{-loop question}}\left(2009\right) } \,,$$which might have some value, like... I dunno. I mean, how much rep would it take to entice @JonSkeet to answer a relatively simple question on an average day – I'd bet he'd do it today for a trillion-rep, if for no other reason than because it'd be funny, but we'd have to ask him!

But say that we look at the inflation of StackExchange rep with respect to how many downvotes it'll buy. Then, in the context of non-Meta answers, that cost has been constantly $$1\,\text{rep}$$ for a while, so it'd be something like $$\left[\text{inflation factor}\right]_{\text{?} \to 2019} ~ = ~ \frac{ 1\,\text{rep} }{ 1\,\text{rep} } ~ = ~ 1 \,.$$ Or we can look at the inflation of the cost to view the website's homepage, which is (and I assume always has been) $$0\,\text{rep} .$$ So then,$$\left[\text{inflation factor}\right]_{\text{?} \to 2019} ~ = ~ \frac{ 0\,\text{rep} }{ 0\,\text{rep} } ~ = ~ \left[\text{indeterminate}\right] \,,$$which is indeterminate because the definition doesn't clarify how to resolve $$\frac{0}{0} .$$

### Conclusion

To sum this up:

1. Inflation is relative to a market basket.

2. Unlike in casual conversations about the US dollar, StackExchange rep doesn't have a standard market basket that someone'll assume you to be talking about if you make a claim about it.

3. So, you can make claims about StackExchange rep having inflation/deflation, but since it's subjective to an unassumed market basket, you'd need to clarify what the reference basket is for the claim to make sense.

• Fun fact: In order to link @JonSkeet's actual user profile, you have to pass a Will save. If you fail, you must instead link Chuck Norris memes.
– Nat
Apr 16 '19 at 3:32
• Jon Skeet has inflated reputation all on his on. He is John Cena in disguise Apr 16 '19 at 16:25