Short answer: by keeping an abstracted measure of value, without turning that abstraction, which we call currency, into capital.
The claim is that capital currency is indispensable because we use flows of it to gather information about exchange. I argue instead that information about exchange is indispensable, and the abstraction of value that facilitates exchange can be divorced from capital. We can give currency the qualities we want it to have and deny it the defects we deplore.
Gathering data doesn't have to be tied to a manipulable, inequitable commodity like a capital currency. When an exchange happens, we will still record inventories, which is most of what markets track. Additionally we can have a treasure chest of complimentary information to contextualize that data, which markets mostly lack.
Mises argument relies on a too-familiar assumption, often asserted by someone looking to stop a conversation: there is no alternative. He said it long before Thatcher did. According to him, markets and only markets provide relevant information. I believe that better information than any market provides can be gathered by asking people what they want and observing what they actually choose to get.
At first, what people ask for and what they will offer to provide will be unequal. In the first iteration of this system, we will ask for (for example) too much booze and not enough bicycles, and other misalignments. Workers who make these things will respond with what they will commit to producing and how much work it will take. Then, people who asked for a gallon of whisky a month will see the aggregate social cost reflected in a realistically indicative price, and some will revise their request to half a gallon. At the same time, some people will notice the cost of a bicycle is lower than they thought it might be, and decide they want one instead of none. These revised requests go back to the workers in aggregate, and a new set of prices are proposed. At each iteration requests fluctuate less, prices get more stable, and when a threshold of tolerance is reached, we have a democratic, self-managed agreement on how much of what will be produced, and how much abstract representation of effort it will cost to buy each product.
Notice that we do not have to *mandate* that people consume all or only what they requested. Those requests can be smeared out into a general geographic or associative population, so people who overestimated their desire for booze can balance out those who underestimated, and demand will still be met. If real demand gets too far ahead of agreed production plus a scheduled reasonable surplus, a real-time price adjustment can be made, and the data collected can be used predictively in the next round of iterations. The process will get better as successive cycles are completed. All the data anyone could want can be collected and analyzed without ever introducing a capital currency, a central planner, or even very much unmet want or wasted production - far less than market capitalism creates now.
Mises had some useful ideas about making predictions based on past behavior. He also had some terrible ideas, like fascism being a worthwhile tool for managing emergencies - such "emergencies" to include direct democracy and attempts to reduce class disparity. While I think his brand of liberalism has done plenty of harm, I also think derivatives of his work in the study of decision making can be applied to a better system.