I would say that it's not just that they send their salary/rent data and know salary/rent data from others. I think the fact that the price is determined by an

I would say that it's not just that they send their salary/rent data and know salary/rent data from others. I think the fact that the price is determined by an algorithm is a key component. To greater and lesser extents, we know what rents are and we know what salaries are. Companies and landlords might have slightly better or more complete information, but it's not like we're totally clueless. A key point of the RealPage algorithm isn't "this is what rent is in your area." The key point is "this is how much you can charge before you start losing renters to other landlords." The problem is when RealPage becomes a market mover - a company that has so much market influence that it doesn't just participate in the market, but exerts control over the market. If every landlord used RealPage and every landlord obeyed RealPage's suggested pricing, it wouldn't be about telling landlords the market price or even predicting how much they could charge before losing renters to other landlords. It would be about predicting how much they could charge before people exited the housing market (eg. by moving away form the city). Of course, if you're doing that in all the cities, then there isn't even other cities that people can move to. Let's say that RealPage is a simple ML program or even a genetic algorithm. It makes some random moves and starts learning "if I do X with rent, Y happens." "Y" might be the landlord losing 1% of their tenants, but with rents going up 10% they're making more money than they're losing. That's "fine" for some definition of fine that accepts money-maximizing capitalist competition as fine (I'm not really interested in debating the housing market in this comment). Now, the issue is that RealPage's algorithm will quickly learn that it can raise rents much higher if it controls the market in a city or neighborhood. If they tell one landlord raises rent by 30%, that landlord is likely going to lose most of their tenants to other landlords. If they tell every landlord to raise rent by 30%, then tenants are no longer in a competitive market where they can rent from someone else. Essentially, the landlords have formed a cartel. Housing activists don't object to open data about what rents are. In fact, housing activists generally support requiring landlords to register rents with the city. Labor activists don't object to open salary data. In fact, labor activists want open salary data because we know that leads to more negotiating power for workers. The issue with RealPage isn't just that it's supplying and collecting data. It's that it is setting prices that, in some markets, a majority of landlords are following. That means that it's no longer a competitive market and (intentionally or not) they've formed a cartel where the algorithm is basically learning that it has the pricing power of a cartel. It's one thing to say "a software engineer at level X makes $Y-Z." It's another thing if 75% of the companies hiring software engineers used the same SalaryAlgorithm which told them "you can lower your engineer salaries by 15% (because we're telling everyone to lower salaries by 15% and we control 75% of the market)." There's actually been a lot of research more recently on monopsony and oligopsony (https://en.wikipedia.org/wiki/Monopsony https://en.wikipedia.org/wiki/Monopsony). Basically, it's the opposite side of monopoly and oligopoly. With a monopoly, you have one company selling something. With monopsony, you have one company buying something. Often this can be companies hiring people (buying labor) and having an impact on wages. If there's only one company hiring software engineers, they're going to control the wages for software engineers. If there's thousands of companies hiring software engineers, they don't have much control - unless they all agree to use the same algorithm for setting salaries and that algorithm quickly learns that it has monopsony power. I think a key part of an algorithm like RealPage's effectiveness can be that it can achieve monopoly/oligopoly power and then learn how much it can charge based on that power - rather than learning how much it can charge based on market rates in a competitive market where landlords aren't colluding (even if the collusion is unintentional). I'm not even suggesting that RealPage's intent was to form a cartel, but if their algorithm is even the tiniest bit smart, it's going to learn pretty quickly that it can rase rents higher when it has monopoly/oligopoly power.