It seems this is a fairly popular topic on Steem, and I have seen a few misunderstandings regarding how the economics of Steem really work. So this is my attempt to explain the economics behind Steem, as I understand it, to hopefully make it a little clearer for everyone. To be clear: this is just my personal view of things and also it is certainly not investment advice.
Steem could be thought of as a decentralized blockchain-based business, but actually it might be more appropriate to use the analogy of a government. It has expenses and revenue. The expenses include paying block producers to run the infrastructure powering the Steem blockchain, liquidity rewards (although that will soon be temporarily disabled with the upcoming 0.12.0 hardfork until a better system can be designed), and of course the major cost of paying out content/creation rewards. Its revenue, much like a government, comes from taxes on its constituents. Specifically, Steem has two types of wealth taxes (hidden as an inflation tax). It taxes liquid STEEM at approximately 50% per year, and it taxes Steem Power at a variable rate of less than 5% per year (these calculations ignore the STEEM supply change effects caused by Steem Dollar conversions at different prices than that which they were issued at; they are also based off the long-term numbers that go into effect after the first year, since the supply inflation rate of STEEM is actually higher than 100% APR during the Steem platform's first year).
If we use the business analogy rather than the government analogy, it doesn't make too much sense to think of taxes as revenue. But one could still relate the tax on Steem Power to a sort of subscription fee charged by the platform which price discriminates users based on the stake the user has in the system (although a larger stake in the platform does provide greater benefits in using the platform's services, such as larger weight in one's votes, a greater curation reward earning potential, and a larger bandwidth quota, so technically it may not be fair to call it price discrimination).
If someone wants to maintain some dollar value of their Steem Power holdings (assuming the market capitalization of STEEM in terms of dollars does not change), they would have to earn or buy more STEEM over time. Some may earn the necessary amount of STEEM to cover the deficit (either in the form of Steem Power directly and/or in Steem Dollars which are then converted into Steem Power) through upvotes on their content or through curation rewards. But many may need to make up for the deficit by purchasing more STEEM from an external exchange using external currency (typically via bitcoin but hopefully later using US dollars directly). They may of course also earn Steem Dollars by trading goods and services they provide and then converting those Steem Dollars earned to Steem Power via Steem's internal exchange.
Now if the market cap rises by at least 5% per year, the users do not need to earn or buy more STEEM in order to maintain at least the same dollar value of their Steem Power holdings (and of course if the market cap rises much more than 5% per year, which in these early stages of the platform does not seem at all unreasonable especially given the recent price performance of STEEM, then users could potentially be earning a decent dollar return on their Steem Power without even needing to earn more through curation or upvoted content). Regardless of the change in market cap, the user still will have a decrease in their fractional ownership of Steem Power (assuming more than 90% of STEEM remains locked up as Steem Power, which is a very safe bet) unless they earn or buy more to offset that decrease. A decrease in fractional ownership of Steem Power for a user means that if everyone else were to maintain at least the same fractional ownership of Steem Power then, assuming all other voting behavior remaining equal, the user would see a reduction in their voting influence relative to others and a corresponding decrease in curation rewards relative to others. It also means that they would get less priority in bandwidth usage compared to others, but it is likely the user would never see this effect in practice if the market cap is rising, because a larger market cap means more funds available to block producers to upgrade their backend infrastructure so that the pie of total available bandwidth grows faster than the shrinking rate of the slice of that growing pie that the user can claim.
If the user does not earn or buy enough Steem Power, then their Steem Power will gradually decrease (as a fraction of all the Steem Power existing within the system) and eventually (even if it takes waiting until market saturation of the platform) they will no longer be able to use the platform (as in post, comment, vote, transfer money, or anything other than only read the content on the platform). So to continue using the platform, they must maintain a sufficient fraction of their Steem Power holdings. Fortunately using the platform at its most basic level without terribly restrictive bandwidth quota limits does not require the user to hold a very large amount of Steem Power in dollar terms. Nevertheless, they do need to continue to maintain at least that minimum amount of Steem Power to be able to reasonably use the system, which means that if the user cannot earn (e.g. through content and curation rewards) that minimum amount to counteract the less than 5% per year effective tax on their Steem Power and thus maintain economic status quo, they will need to buy more STEEM (and power up) every year. This effectively amounts to a subscription fee to use the platform. And that is the reason why the platform could theoretically continue to operate indefinitely even after reaching market saturation. Now whether it actually does so or not depends on many factors, but I think the most critical of those factors is whether people would be willing to pay that effective subscription fee (in the form of a less than 5% wealth tax on their Steem Power) to continue using the platform.
Now I do think there are good reasons why people would be more willing to pay that "effective subscription fee" to use Steem's social media platform even when it is so hard to get users to pay money to use any other social media or social networking platform. First, Steem doesn't nickel-and-dime users. I think the cognitive burden added to users if every post, comment, and vote cost them a small micro-transaction fee would be enough to dissuade them from using the platform. Users would likely find it more palatable to pay even a larger total amount of money through a monthly subscription fee than being charged for every tiny interaction with the platform. Second, even though an inflation tax is basically just a hidden wealth tax (of course a tax only on the specific asset being inflated), the former is psychologically more acceptable because users do not like to see the numbers on their screen decrease. And third, and most importantly, I think the content/curation reward incentives really change the game as far as enticing a user to pay a subscription fee to access the platform. Psychologically, I think it can work for the same reason people still play the lottery or casino games, despite the wide availability of knowledge on the mathematics behind those games and how on average the user is expected to lose money. And it isn't necessarily the case that people are just being stupid and don't understand the math (although I am sure there is a lot of that as well). I think for many people, deep down they understand the fact that on average they will likely lose some money (unless it is a casino game of both skill and chance, and they actually are much better skilled at the game than their opponents), but as long as they limit the amount they put up for potential loss to something they can tolerate, it is worth it to them for the fun and excitement that they experience.
I believe the same psychological dynamics could take place with Steem's social media platform. The comparison is better made to games requiring some skill, since it is clear that writing ability, deliberate timing of when one posts, and the content one writes about are all very important factors influencing the amount of payout one's post will receive. But there is some element of randomness that is involved as well. Some posts that many people would expect to get a high payout do not get all that much. And some posts many people wouldn't expect to get much money end up getting tens of thousands of dollars worth of payout. These are similar dynamics to playing many casino games which are clearly appealing to many people. And whether one wins big or doesn't win at all, similar to the casino games providing them joy and excitement from simply playing the game, Steem still provides value even with no payouts because it allows people to communicate with others and find interesting content on a decentralized censorship-resistant social media platform. But it is due to the psychological dynamics similar to the casino games that I believe the Steem social media platform has some chance of doing what is incredibly hard to do for social media platforms: getting users to pay to use it (other than through advertising of course).
If Steem just had a flat pay subscription model to use the social media platform (which with the exception of being decentralized and maybe allowing optional tipping would be no different than existing social media or social network platforms), I don't think people (other than a very tiny niche community who really strongly values decentralization) would be willing to pay the costs to use it when they could instead use Reddit or Twitter for free (free meaning ad-supported of course). But because of the gamification and the very real possibility of earning large payouts that can actually be used (perhaps after a few steps of conversion) to buy real goods and services, I think it changes the psychology of being willing to pay to use the platform entirely. And if it does succeed in doing so, then there can be a continual source of revenue (buy demand for STEEM to power up to just maintain the same economic value of Steem Power and thus maintain the expected bandwidth quota status quo) to keep the platform operating even when it reaches saturation. But of course no one really knows what the future will hold. It is important for people to remember that this is all a brand new experiment and thus of course a large financial risk. I think the advice "don't invest more than you can afford to lose" is generally good advice and I think it certainly applies in the case of Steem as well.