Slashdot Mirror

← Back to Stories (view on slashdot.org)

Left To Their Own Devices, Pricing Algorithms Resort To Collusion (popularmechanics.com)

Posted by msmash on from the closer-look dept.

Reader schwit1 shares a report: When you're browsing online, who sets the prices? An algorithm, most likely. A study from 2015 showed that a third of all items on Amazon [PDF] had prices set by an algorithm, and chances are that percentage has only risen. A new study shows how easy it would be for price-setting algorithms to learn to collude with each other and keep prices at a disadvantage for customers.

This sort of collusion would stem from a certain type of algorithm, the researchers say. Reinforcement algorithms learn through trial and error. In the simplest terms, a walking robot would take a step, fall, and try again. These algorithms have often been used to teach algorithms to win games like Go.

"From the antitrust standpoint," say professors Emilio Calvano, Giacomo Calzolari, and others from the University of Bologna in Italy, "the concern is that these autonomous pricing algorithms may independently discover that if they are to make the highest possible profit, they should avoid price wars. That is, they may learn to collude even if they have not been specifically instructed to do so, and even if they do not communicate with one another."

20 of 128 comments (clear)

  1. Collusion? by XanC · · Score: 3, Interesting

    "and even if they do not communicate with one another"

    Well that can be the case with people, too, can it not? Except that when people do not communicate with each other, I don't believe that it can be said to be colluding, by definition. So why is it collusion when algorithms do it?

    1. Re:Collusion? by Shaitan · · Score: 4, Insightful

      "Independent algorithms (or people) arriving at the same conclusion from the same data and openly performing actions in their own self-interest without communication is not collusion."

      If it comes to the same anti-consumer result and the same result as monopoly behavior in a free market it is collusion. Or insert a new word for what is bad and can't be tolerated if we are to have a sound economy.

      It doesn't matter if it is one monopoly fixing terms and pricing in an anti-competitive way, 8 companies with common interest doing the same, or 10,000 doing the same. The active communication isn't the issue. And it isn't just pricing, the major chip manufacturers engage in this sort of anti-competative behavior and coordinate on release rates by releasing roadmaps of how far they are planning to go with their capacities and features in advance via unspoken (or possibly spoken behind closed doors) agreement with each other.

      "Collusion is an agreement between two or more parties, sometimes illegal–but always secretive–to limit open competition by deceiving, misleading, or defrauding others of their legal rights, or to obtain an objective forbidden by law typically by defrauding or gaining an unfair market advantage. It is an agreement among firms or individuals to divide a market, set prices, limit production or limit opportunities."

      An unspoken agreement fits that definition just fine.

    2. Re:Collusion? by Layzej · · Score: 3, Interesting

      "and even if they do not communicate with one another"

      Well that can be the case with people, too, can it not? Except that when people do not communicate with each other, I don't believe that it can be said to be colluding, by definition. So why is it collusion when algorithms do it?

      In most countries (including Europe and the US) such ‘tacit’ collusion, not relying on explicit intent and communication, is not currently treated as illegal, on the grounds that it is unlikely to occur among human agents and that, even if it did occur, it would be next to impossible to detect. The conventional wisdom, then, is that aggressive antitrust enforcement would be likely to produce many false positives (i.e. condemning innocent conduct), while tolerant policy would result in relatively few false negatives (i.e. excusing anticompetitive conduct). With the advent of AI pricing, however, the concern is that the balance between the two types of error might be altered.

    3. Re:Collusion? by ezdiy · · Score: 2

      Perfect markets produce "cartel" like this when all participants have near equal production capacity - they'll all gauge the price to maximize profit and minimize output as that is the most optimal nash game. And they can do it with no direct collusion, just market ticker.

      However whether amazon is a perfect market with participants of comparable capacity begs question. Real world markets have a stream of newcomers who pull down the majority to their price level and would even refuse to participate in a cartel which would not guarantee same market share they'd have otherwise (see OPEC politics).

      Seems like something far simpler is at play - amazon promoting more expensive resellers in search results, because they'll get more of a cut from those.

    4. Re:Collusion? by ljw1004 · · Score: 2

      If two algorithms are setting prices and learning from how each other is setting prices, then they are communicating. It's a fairly low bitrate communication but it's certainly communication.

      As for intent? I don't know to use that word in relation to algorithms.

    5. Re:Collusion? by CohibaVancouver · · Score: 2

      But they are communicating via the prices they set.

      Exactly, Anonymous Coward.

      If you have a hotdog stand selling smokies for $5 and I set up a hotdog stand two blocks away, note your price, and also sell *my* smokies for $5, I'm not "colluding" - I'm just price-matching. Seems to me that's what these price-bots are doing.

    6. Re:Collusion? by mpercy · · Score: 2

      Prices do indeed constitute communication, but they are in the open. Something done openly is not collusion.

    7. Re:Collusion? by swillden · · Score: 3, Insightful

      Prices do indeed constitute communication, but they are in the open. Something done openly is not collusion.

      That's not true. Collusion is defined as "secret or illegal cooperation". It generally is secret because doing illegal things openly is likely to result in getting caught, but that doesn't mean that collusion can never be done in public.

      --
      Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
    8. Re:Collusion? by larryjoe · · Score: 2

      Perfect markets produce "cartel" like this when all participants have near equal production capacity - they'll all gauge the price to maximize profit and minimize output as that is the most optimal nash game. And they can do it with no direct collusion, just market ticker.

      Perfect, working markets will use public prices as signals to adjust prices and capacity. This is the way supply and demand is supposed to work, and it can be done without direct collusion. However, there are situations where this dynamic adjustment of prices doesn't work, e.g., if there is a monopoly or if there are no disruptors (i.e., all vendors are willing to keep their market share without competing).

      The problem with the robo-pricers is not the computer aspect. It's that there are no disruptors in the market, either because no one cares to be a disruptor or because the barriers to entry as a vendor are too great.

    9. Re:Collusion? by budsetr · · Score: 2

      Posting your prices where your competitor (or anyone) can see it IS communication.

    10. Re:Collusion? by dryeo · · Score: 5, Insightful

      It's a good example of how the market does not always result in the cheapest prices.

      --
      https://en.wikipedia.org/wiki/Inverted_totalitarianism
    11. Re:Collusion? by RhettLivingston · · Score: 2

      I had the same thought. It seems that many businesses are no longer doing their best to put their competition out of business in order to increase profits. Instead, they concentrate on ways to increase profits through internal changes without changes in sales. Many also no longer push to innovate. R&D cost money. I guess it is much easier to effect the stock on 3-month timeframes when you just silently agree to a truce with your competition and work on reducing internal costs and using accounting tricks to increase profit.

  2. Collusion? by mpercy · · Score: 5, Insightful

    "That is, they may learn to collude even if they have not been specifically instructed to do so, and even if they do not communicate with one another."

    I think collusion pretty much requires communication and intent.

    Independent algorithms (or people) arriving at the same conclusion from the same data and openly performing actions in their own self-interest without communication is not collusion.

    Their actions may look like collusion from the outside, but they fail on most commonly accepted definitions of the word.

    E.g. wikipedia

    "Collusion is an agreement between two or more parties, sometimes illegal–but always secretive–to limit open competition by deceiving, misleading, or defrauding others of their legal rights, or to obtain an objective forbidden by law typically by defrauding or gaining an unfair market advantage. It is an agreement among firms or individuals to divide a market, set prices, limit production or limit opportunities."

  3. How is that different? by Shaitan · · Score: 2

    "That is, they may learn to collude even if they have not been specifically instructed to do so, and even if they do not communicate with one another."

    Sounds like the situation we have with major companies now.

  4. the 1970s called... by goombah99 · · Score: 5, Informative

    There's case studies many economics classes use of how the airlines in the 1970s would evade the collusion in restraint of trade laws. Basically, they developed a process for proposing fare changes. They would announce a scheduled fare increase a month in advance. If the other company followed suit they would enact it if they didn't they would retract it or post a fare decrease shortly after that.
    It worked quite well till the regulators figured it out. Braniff (deceased high end airline) was the ringleader.

    These pricing algorithms just accomplish this one time scales of hours rather than months.

    but it's just old wine in new bottles. No one invented anything or changed anything. But the amplification and universality of it made the problem worse.

    But what actually makes this interesting is that it may also be an emergent behavior as opposed to either intentional programming or an untended artifact of some algorithm. THat is, if an AI is simply asked to maximize profit in a multi-agent system it's entirely possible it will learn this tit for tat strategy. Humans do in cooperation-games in behavior theory.

    --
    Some drink at the fountain of knowledge. Others just gargle.
    1. Re:the 1970s called... by logicnazi · · Score: 2

      Actually, I suspect it isn't even worse. The airline situation in the 70s was discovered and proved because it was so obvious.

      For instance there are a number of studies now showing the cross ownership (one mutual fund owning many companies in a market) substantially decreases competition even when there is never any easily observed collusion or signalling like you describe.

      This suggests humans have a wide range of subtle communications channels they can use and don't even need to be aware that they are colluding to accomplish the same results (they can just be thinking hey, it would be great if we could raise rates let me try and guess if my competitor would undercut me if i did so). So I strongly suspect that such price collusion is already happening and algorithms make it no worse.

      --

      If you liked this thought maybe you would find my blog nice too:

  5. Wrong word by jbmartin6 · · Score: 4, Interesting

    That is called equilibrium, not collusion. There once was a similar state in the US with breakfast cereals for a while, until one of the participants decided to break the equilibrium and lower prices to try to gain market share.

    --
    This posting is provided 'AS IS' without warranty of any kind, implied or otherwise.
  6. Re:Someone doesn't know the definition of COLLUSIO by viperidaenz · · Score: 2

    The algorithms are communicating with each other through the prices they constantly alter.

  7. Re:Collusion or not by slacktide · · Score: 3, Informative

    No, it's pretty obvious that Amazon DOES NOT adjust adjust pricing on a per-consumer basis, and they do not use your past purchasing history to set pricing specific to you. If they did, then sites like camelcamelcamel would not work. Amazon does adjust pricing over time, but at any given time, all buyers are offered the same price.

  8. No, it's not equilibrium by BankRobberMBA · · Score: 2

    Equilibrium is when disparate forces in a market reach a balance point, not when 'competitors' refuse to compete. The cereal problem you mentioned was a classic case of refusing to compete. FTC watched those guys for YEARS trying to catch them communicating.

    An example of an equilibrium is when prices go low enough that it becomes uneconomical for some participants to stay in the market if prices go any lower, but there is also not enough profit in the market to entice any new entrant (who faces barriers to entry), so prices and market participants stabilize. This is usually a sign of strong competition.

    The diagnostic symptom of a cartel (either communicating or not) is sellers making economic profits (as opposed to accounting profits) for a sustained period of time. Economic profits are profits after deducting opportunity costs. If it's more than one company, you are looking at a cartel. If it's only one company, you're looking at a monopoly.