X is offering millions of dollars in rewards for good articles. What kind of content is getting paid?

Author: David, Deep Tide TechFlow

In mid-January, X announced a $1 million reward for the best-performing long-form articles on the platform.

Elon Musk personally retweeted and confirmed it. The rules are simple: US users only, original English articles of 1000 words or more, ranked primarily by impressions from US paid users.

You might recall that a few days before this content incentive campaign was launched, personal growth blogger Dan Koe published an article titled "How to fix your entire life in 1 day," which garnered 170 million views and became X's best-performing article ever.

X clearly saw the traffic potential of long articles and quickly followed suit; lowering the threshold for the Articles feature, adjusting the algorithm weight to prioritize long articles over short posts, and announcing a million-dollar essay contest prize.

The competition lasted two weeks and attracted tens of thousands of participants.

The results were announced on February 4th, with a final total prize pool of $2.15 million, more than double the promised amount. The champion received $1 million, the runner-up $500,000, plus a $250,000 "Creator's Choice" award and four honorable mentions of $100,000 each.

The awards are roughly as follows:

You can see that Dan Koe is on the list again. However, his previous article on how to fix your life in one day had 170 million views, while the winner of this writing contest only got 45 million.

Viral hits are still rare, but it's worth analyzing some of the award-winning articles.

Champion: A "smurf account" with 90,000 followers won 1 million yuan using a self-built database.

The title of the article by the champion @beaverd translates to "Deloitte, a $74 billion cancer spreading across the United States." It's about the well-known consulting firm Deloitte.

This account currently has "only" 90,000 followers, which is relatively small compared to the other winners, and it has no endorsement from any media organization or other sources other than a verified blue V badge.

His article didn't use any trending keywords, but the things he revealed were quite sensational: how Deloitte took $74 billion in contracts from federal and state governments and then ruined the projects.

Portal here

If you click in, you'll find that this person has really put in the effort.

He created a website called somaliscan.com, scraped millions of government invoices, and cross-referenced them one by one with audit reports and system failure records.

Then, using this firsthand data, a series of shocking stories were told: California's unemployment benefits system was defrauded of $32 billion, Tennessee's Medicaid system collapsed, leaving 250,000 children without protection, and the court's IT upgrade burned through $1.9 billion and remained unfinished... covering a total of 25 states.

He also uncovered the revolving door between Deloitte executives and government officials, detailing who jumped from Deloitte to which department, and which contracts they approved and returned to, with names and amounts clearly listed.

One person built his own database and earned $1 million through self-study.

Runner-up: A financial blogger with 700,000 followers teaches you how to make money amidst tariff panic.

The runner-up, @KobeissiLetter, is a familiar face in the macroeconomics and finance circle, with 700,000 followers, and has long followed US economic policies and market fluctuations.

His article was also very straightforward, breaking down Trump's usual tactics of using tariffs into a repeatable trading framework, titled "Trump's Tariff Script: An Operational Guide".

Because Trump often acts unpredictably, likes to issue outrageous policies and threaten other countries, but doesn't always follow through, some on Wall Street have summarized this pattern as TACO, which stands for Trump Always Chickens Out.

TACO refers to a recurring pattern:

Trump announces harsh tariffs → Market crashes → A few days later he softens his stance or postpones the tariffs → Market rebounds.

Portal

What KobeissiLetter did was transform TACO from a joke into a time-based trading manual. He used the tariff events of the past 12 months as a sample to create a complete cycle template for you to trade according to the time period.

For example, the White House releases information over the weekend to create panic, funds rush in to buy at the bottom during the week, signals of easing comebacks are given the following weekend, and some kind of agreement is reached within 2 to 4 weeks. At the same time, he will continue to post updates as each step is fulfilled, telling you where you are now, making it more like a pre-production research series.

He also offered a more practical approach: monitor the yield on the 10-year US Treasury bond. If this figure exceeds 4.60%, Trump will likely back down.

This kind of thing is perfect for X's paid users who focus on macroeconomics and trading.

It doesn't discuss whether tariffs are good or bad, nor does it make moral judgments. It just tells you what actions you should take at what time to make money next time this happens.

Third Place: DAN KOE, the most liked, with a familiar philosophy of life.

Dan Koe's entry, "How to Enter a State of Extreme Focus Anytime," received 42,000 likes and 8,681 shares, the highest among all entries. However, its exposure was only 11.04 million, less than a quarter of the winner's.

The prize X awarded him wasn't strictly third place; it was a separate "Creator Choice" award worth $250,000.

It's understandable, really. Dan Koe was the one who "inspired this game." His viral article in early January, which garnered 170 million views, showed X just how high the ceiling for long-form articles can be.

Portal

I won't go into too much detail about the article itself; it's basically the same old methodology for personal growth. It mainly discusses how to gain focus, using concepts from neuroscience and flow states to support and elaborate on these points.

Actually, this article had the best interaction data, but according to the core rule of the competition, "exposure to paid users in the United States," it wouldn't rank highly.

Why do the most engaging articles not get much exposure? We'll discuss this discrepancy later.

Honorable Mention: 100,000 x 4

Nick Shirley, Josh Wolfe, Kaizen Asiedu, and Ryan Hall each received a $100,000 incentive. Their accounts cover four areas: public policy, geopolitics, history, and public safety.

Josh Wolfe, co-founder of Lux Capital and a well-known venture capitalist, also announced that he would donate the entire prize money to four charities.

Since the original post did not list the specific articles by these four individuals, and due to time and energy constraints, we were unable to conduct further investigation. We welcome everyone to supplement the information.

Some in-depth observations

Some patterns can be observed from the results of this competition:

• The article with the most likes only received a quarter of the exposure of the winning article.

The most counterintuitive statistic in this match is definitely Dan Koe's.

With 42,000 likes, 8,681 retweets, and 4,627 comments, it had the highest engagement across all categories. However, its exposure was only 11.04 million, less than a quarter of the winner @beaverd's. @beaverd, on the other hand, had 30,000 likes, fewer than Dan Koe.

If you've ever worked in social media marketing, this data might seem odd. Generally, higher engagement means a stronger algorithm that's more likely to promote your content, resulting in greater exposure.

However, X's competition this time doesn't calculate total impressions, but rather "US paid user homepage timeline impressions." This metric excludes all visits from non-US users, non-paid users, search engines, and personal homepages.

Dan Koe writes about personal growth, naturally attracting a more global audience, with a large number of non-US users among his followers. @beaverd writes about how Deloitte is wasting American taxpayers' money, naturally attracting an audience concentrated in the US. Under the same algorithmic recommendation mechanism, the "geographic concentration" of the content determines the level of this metric.

• 90,000 followers beat 900,000 followers: content scarcity > follower base

The champion, @beaverd, had 90,000 followers before the competition. The runner-up, @KobeissiLetter, had 700,000 followers. Dan Koe had 900,000 followers.

If the number of followers determines exposure, the ranking should be the other way around. However, actual results show that in X's Articles recommendation logic, the weight of the number of followers is far less than imagined.

@beaverd's victory hinges on whether he possesses something others lack, or whether the scarcity of his content played a crucial role.

This is completely different from the traditional logic of traffic. Big accounts rely on the number of fans and the frequency of posting, but in an environment where algorithm-driven distribution prevails, "whether you have something exclusive" is more important than "how many fans you have".

• You need to build your own content "hardware".

Looking at it from another angle, the topics of these three award-winning articles are completely unrelated: one exposes government contracts, one teaches you how to trade tariff fluctuations, and one talks about how to focus your attention.

In any content platform's classification system, they wouldn't appear on the same list. But they have one thing in common: each piece has its own independent "hardware," in other words, you need a narrative framework.

@beaverd's hardware is a self-built database crawling government data; KobeissiLetter's hardware is a trading framework that has been backtested for 12 months; and Dan Koe's hardware is a six-chapter methodology that integrates neuroscience and psychology. Although it may look profound, it is actually based on principles that everyone knows.

None of the award-winning articles were purely opinion pieces. They all required considerable length to convey a wealth of information, which is precisely the reason for the existence of the X Articles product format.

Another noteworthy fact is that none of the eight winners came from traditional media outlets.

All of them are independent creators. It's not that traditional media didn't participate, but under this competition format, individual accounts actually have an advantage.

Institutional media outlets typically publish their content on their own websites, with social media platforms only including a link and summary. However, Articles requires full content to be published on X site, which is an awkward move for media outlets accustomed to external traffic redirection.

What exactly did X buy with 2.15 million?

Let's return to the platform itself.

X initially promised $1 million in incentives, but ultimately paid out $2.15 million. During the competition, they also implemented a series of supporting measures: expanding the Articles feature from creator accounts to all paying users, adjusting the algorithm to increase the recommendation weight of long-form content, and changing the scoring method to "homepage exposure for US paying users."

The most direct reason for spending such a large sum is that X needs original long-form content on the site.

In the past, long-form content on X relied primarily on external links from Substack, Medium, and personal blogs. Users would click and leave, leaving their reading time and interaction data to others. Articles aims to keep this content on the site, ensuring users read from beginning to end without leaving X.

Going a step further, X has Grok. Training large language models requires high-quality, long-text data, while the vast majority of content on X consists of short tweets of 280 characters. If Articles can consistently attract creators to produce in-depth, long-form articles, this content will serve as training material for Grok.

Finally, the value of paying users.

The competition rules limit the metrics to "exposure on the homepage of paid users in the United States," which is tantamount to telling creators directly that their content must serve paid users.

This is using creators' content to support the paid system, making paying users feel that "the money I spent is worthwhile because I can see in-depth content on the homepage that I can't see elsewhere."

From the perspective of content creators, we believe that the era of pure opinions may be coming to an end.

This trend also applies to creators in the crypto space. The crypto industry is never short of opinions; countless people on X (a crypto platform) are constantly making trading recommendations, predicting prices, and commenting on regulations.

However, very few people can build an on-chain data analysis tool like @beaverd, or break down market cycles into repeatable trading scenarios like KobeissiLetter.

Maintaining scarcity and independence while consistently producing output is actually a highly specialized and rewarding endeavor.

We also hope to see more content from the Chinese-speaking world appear on the list in the future.