Information Value: How to Distinguish Sand from Gold

You spend hours every day scrolling through your phone. You follow all the industry big shots. You listen to hardcore business podcasts at 1.5x speed with noise-canceling headphones on your commute. You’ve joined dozens of WeChat groups. You spend hundreds of dollars a month on AI subscriptions. Because you want to know where the next big thing is, you must be the first to use the latest tools, you need to understand policy trends, and you have to keep up with hot topics…

This sentiment sounds positive, but it’s actually fear—English even has a specific term for it: FOMO (Fear of Missing Out). You’re afraid you haven’t used that most disruptive new model, you’re afraid you won’t understand the buzzwords your colleagues are using, you’re afraid of being left behind by the times.

But after receiving so much information, shouldn’t you become stronger and more composed? Why are you still living in constant anxiety?

Because that information has no value.

“Value of Information (VOI)” is a specialized theory in decision science. I hope it can help you distinguish sand from gold.

✵

VOI theory originated in the mid-20th century, representing a paradigm shift in statistics and management science [1]. Previously, scholars focused on judging the truth or falsehood of information and how to acquire it. However, true information isn’t necessarily useful to you. VOI theory was the first to care about the value of information, and its definition of value is very strict:

Information has value only when it can change your actual actions.

Simply put, it has value only if it’s useful for decision-making.

VOI is the difference in the expected profit between the best choice you can make with the information and the best choice you could make without it.

A classic academic case goes like this. Imagine you are the CEO of an oil company. There’s a piece of land in front of you. The cost to drill an oil well here is $10 million. If there’s oil underground, the completed well is worth $50 million, resulting in a net profit of $40 million. If there’s no oil, the $10 million is wasted. Geologists analyze the land and say the probability of oil is 20%.

Now a surveying company approaches you, saying they can perform a precise seismic scan to tell you exactly whether there’s oil. What’s the maximum you’d be willing to pay for this report? VOI theory requires us to calculate the expected return first.

If you choose to drill now without the report, you either lose $10 million or gain $40 million. The expected return is 20% × $40 million - 80% × $10 million = 0. So the rational decision is not to drill. With the report, if it says “Oil,” you drill and make a steady $40 million; if it says “No oil,” you don’t drill and lose nothing. Since the prior probability of “Oil” is 20%, the expected value of the report is 20% × $40 million = $8 million.

VOI theory says that increasing your return from $0 to $8 million is the value of this information.

Thinking more about this oil company example will give you more composure when facing new information. Their report is worth $8 million, whereas that “10,000-word analysis of the AI revolution” you read last night might have a VOI of 0.

Information without a decision target has a value close to zero.

✵

You might argue, “Isn’t lifelong learning important? Isn’t it about empowering oneself? Isn’t there ‘utility in the useless’?”

From the perspective of VOI theory, valuable information must be related to a decision you need to make—either now or in the future. Of course, you can spend your attention on information unrelated to decisions, but that is “consumption value,” not “decision value.”

Think about those hot topics that keep us in a state of FOMO all day. Most probably have nothing to do with your decisions. You are actually seeking a sense of self-satisfaction—“I’m getting stronger, I’m keeping up”—or confirming your identity: I’m someone in the know, someone at the forefront of the times, and preferably someone who knows earlier than others.

People often equate consuming information with taking action, but they are two completely different things. Only information that can change action is valuable.

Information anxiety doesn’t come from a lack of information, but from a lack of decision execution. To use the words from our discussion on WOOP, those who consume information all day without using it to make substantial decisions are actually in a state of “drifting”—thinking of a thousand paths at night but grinding tofu as usual in the morning. To break out of the drifting state, you must have a plan, an executable action.

Value theory sounds dry, but it’s actually quite exciting because it tells you what the “good stuff” is.

✵

From a VOI perspective, high-value information often has three characteristics.

First, it is goal-oriented. You must first want to do something and have a goal before you can talk about chasing high-value information. What exactly do you want to optimize? What is the bottleneck of what you are doing now? Good information should be a lever; it can change your actions and have a practical driving effect on your goals.

Such information is often local and specific to you, rather than trending news. What the Japanese Prime Minister said probably has nothing to do with you. You should care about the actual productivity distribution of your team: who is really carrying the weight, who is the bottleneck, and who is just looking busy.

Second, it carries “pain.” Our brains all have “Confirmation Bias”; we like to see information that confirms our original views. But truly valuable information must be information that does not fit your previous cognitive model; it will change your priors.\

This kind of information will not only surprise you but often make you uncomfortable. It might be negative feedback or something that disproves your ideas.

Third, it appears on the “decision boundary.” Only when you are swaying between two options—undecided whether to choose this or that—and a key piece of information makes your choice tip immediately, is it high VOI.

For example, if you’ve already decided to buy a certain electric vehicle, the money is ready, and you’re just waiting for the weekend to pull the trigger. If you see another analysis report on electric vehicles today, you’ll be interested, but it won’t actually sway your decision—that’s low VOI. Conversely, if you have two job offers in hand and are struggling to decide because each has pros and cons, and a friend suddenly reveals a secret: one of the companies is preparing to go public within a year or two! That news is worth a thousand pieces of gold to you.

In contrast, the information we FOMO about every day is: 1. accessible to everyone, 2. mainly provides emotional value, and 3. is basically unrelated to your decisions. Therefore, the VOI is very low.

Let me give you a real-world example [2]. There is a prediction market called Polymarket where you can bet on various public events. Between January and February 2026, a PhD student made about three thousand bets on major sporting events and profited nearly $3 million. How did he do it?

Prediction markets provide odds before a game. Those odds are carefully calculated and basically reflect the true strength of both teams [3]. This guy discovered that the Asian betting markets update their odds two to three hours earlier than Polymarket. In other words, during those two to three hours, the Asian market has newer and potentially more accurate predictions than Polymarket.

This is equivalent to someone telling you when you’re undecided: “A professional just made a judgment based on the latest situation, and the probability of that team winning has increased slightly.”

This “slight” difference is enough to tip the scales. Of course, if you only do this once or twice, you’ll win some and lose some—but if you do it in large volumes, that tiny probability difference allows you to profit steadily on average. This guy used a betting bot to scrape Asian markets and Polymarket odds in real-time and automatically place bets on Polymarket based on the differences—basically using a system to make money.

What was the high VOI here? It wasn’t that the betting market had these projects, nor the odds themselves, but the difference in odds between the two markets. No one would specifically write down that difference to tell you. And it was fleeting. We can imagine that if many people imitated this play, the difference between the two markets would level out, and the arbitrage space would disappear.

✵

To acquire high-value information, you should be an information sniper, not an information wanderer. You must first think about what decision you want to make, whether you are currently near or far from the decision boundary, whether this new information could significantly impact your decision, and then you have to look at its cost.

VOI theory does not advocate for us to acquire all relevant information. You don’t need to keep researching; sometimes you just have to make a call. Information has strongly diminishing marginal utility, and knowing when to stop researching is also a skill [4].

However, there are more advanced operations here. For high VOI information, sometimes you just have to be willing to pay the cost. We need to understand two concepts: EVPI and EVSI [5].

EVPI is the “Expected Value of Perfect Information.” It completely eliminates the uncertainty of the matter, like information from a god’s-eye view. The seismic scanning service for oil mentioned earlier provides EVPI because it tells you exactly whether there is oil. Another example is if you want to buy a used car and pay a professional mechanic to give it a thorough inspection—that’s acquiring EVPI.

However, perfect information is often unattainable, which leads to EVSI, or the “Expected Value of Sample Information.” EVSI is “partial information” you acquire through experiments, sampling surveys, or active intervention. Although EVSI retains some uncertainty, it is still very valuable as long as it makes your probabilities more accurate and updates your priors.

Conducting clinical trials for new drugs, running A/B tests for internet companies, or launching a Minimum Viable Product (MVP) to see market feedback are all ways to acquire EVSI.

These principles are simple. Let’s look at what high-value information looks like in daily life—there isn’t actually much.\

✵

Ordinary employees in the workplace often overestimate the value of macro industry trends. They like to talk about “windows of opportunity” and care about financing news from industry giants—actually, those are things only top management needs to worry about, or they are just “pies in the sky” painted by the boss. What you should care about most is: What is the specific reward function of your company’s field? What are the core metrics for promotion? Which step in your current workflow is the most stuck? Is there room for optimization?

Students often overestimate information about further education and famous schools, such as “Predictions of the hottest majors for the next decade” or “Schedules of top students at Tsinghua.” Actually, you should care more about where your weak points are: Which concepts don’t you understand? Which types of questions do you always lose points on? Which are the points you should have secured? Get your grades up first, then consider how to choose your major based on your specific situation.

Some researchers like to talk about national affairs and recent world-shaking scientific discoveries, but those are of little use. Most researchers prefer to talk about “academic gossip”—for example, which professor has a conflict with another, which big name was just elected to the academy, and so on. This won’t help you either. What you should really care about is: What are the current hot problems and mainstream approaches in your niche field? Why can’t you get your own experiment to work? What is the noise floor of your instrument? Why did the research group next door get a large grant?

Most interesting are the business owners. They love going to business schools to talk about trendy concepts like “the second curve” or “disruptive innovation,” and some even study the I Ching and metaphysics. But what they really should know is: Why did your best salesperson quit last month? What exactly is going wrong internally with that supplier who is always two days late? You need to figure out constraints, feedback, and causal relationships.

And so on. Simply put, various public reports, grand narratives, and macro trend predictions are all overestimated information—they look gorgeous but are actually sand.

High VOI information is often fragmented, unglamorous, or even sounds a bit boring; it will never appear in media headlines. It might be some dry prior probabilities, negative feedback from users, internal information revealed by “structural holes” between two circles, or, most importantly, your own data.

✵

Some people talk eloquently about international situations but are dismissive of their own organization’s politics; some track AI revolution news every day but haven’t even implemented a simple automated workflow; some will compare three different stores and read all the reviews for a small item worth a few dollars online but have never done due diligence for major life decisions like buying a house, changing jobs, or even getting married…

Some people greedily absorb knowledge just to satisfy their curiosity—I highly respect curiosity—but some collect information for decision-making.

It’s fine to be an “information person,” but if you want to get something done, you must have VOI awareness.

【Concluding Poem】

Set options first, then seek the lead. Action unchanged is waste indeed. Why know it all? Anxiety’s snare. Beyond the pivot, just idle air.

Notes

[1] Howard, Ronald A. “Information Value Theory.” IEEE Transactions on Systems Science and Cybernetics 2, no. 1 (1966): 22–26.

[2] See: X tweet by Blockchain Market Research (@qkl2058), February 17, 2026. https://x.com/qkl2058/status/2023757218932552134?s=12&t=cBYTMvCM9D0ac23JcB2osA The protagonist’s account page is at https://polymarket.com/@432614799197\

[3] Elite Daily Lessons Season 2, “How Old Fans Bet Scientifically.”

[4] Pirolli, Peter, and Stuart Card. “Information Foraging.” Psychological Review 106, no. 4 (1999): 643–675.

[5] Raiffa, Howard, and Robert Schlaifer. Applied Statistical Decision Theory. 1961.