The Downside of MEV

Why Thousands of Trading Bots Continue to Compete in the Race for Profits

Over the past few years, MEV has been considered one of the most profitable areas of the crypto industry. Stories about bots that have made millions of dollars through arbitrage and front-running have long been part of crypto folklore. It seems that all you need to do is write a quick algorithm—and profit becomes merely a matter of speed.

A new study titled “There Will Be Spam: Characterizing State-Invariant Transactions and Speculative MEV,” authored by researchers Vabuk Pahari, Johnnatan Messias, and Christof Ferreira Torres, offers a fresh perspective on MEV for the first time.

Instead of analyzing the most profitable deals, the authors decided to find out how much the race for that profit costs the industry.

To do this, they set up archive nodes for Ethereum, Optimism, and Base, analyzed nearly two years of network history, and processed hundreds of millions of transactions.

The results turned out to be unexpected.

The study shows that a significant portion of the activity on modern blockchains is generated not by ordinary users at all. When the cost of all failed attempts is taken into account, many speculative MEV strategies on Layer 2 networks turn out to be significantly less profitable than is commonly believed.

In other words, for one algorithm to be successful, hundreds of others must first fail, pay fees, and create network congestion.

It was precisely this flip side of MEV that turned out to be the study’s main finding.

---

Nearly half of all activity may not create value

In the course of their research, the authors identified a separate category of operations: State-Invariant Transactions (SIT).

These are transactions that:

• take up space in the block;
• consume computing resources;
• pay a commission;
• but have virtually no effect on the state of the blockchain.

The results of the analysis were unexpected.

Network	The Proportion of State-Invariant Transactions
Ethereum	2,6%
Optimism	24%
Base	37%

After the Dencun upgrade, the situation became even more interesting. Throughout much of the period under review, State-Invariant Transactions accounted for about 40–50% of all daily transactions on Layer 2 networks and consumed a similar amount of gas.

---

Where do these transactions come from?

In the next step, the researchers attempted to determine the origin of SIT.

The result turned out to be unexpected.

Network	SITs related to speculative MEV
Ethereum	≈3%
Optimism	≈73%
Base	≈60%

For Ethereum, the majority of such transactions consisted of failed smart contract calls and repeated user actions.

In Layer 2, the situation turned out to be different—speculative MEV strategies became the main source.

---

Numbers That Are Changing Our Perception of Profitability

The study found that:

Ethereum

• about 1,845 MEV bots;
• approximately 1.6 million successful MEV transactions;
• approximately 670,000 state-invariant transactions;
• The overall success rate is about 72%.

Optimism

• approximately 4.3 million successful MEV transactions;
• nearly 89 million State-Invariant Transactions;
• The overall success rate is about 4.8%.

This means that for every successful operation, there were approximately 20 failed attempts.

Base

• approximately 14.4 million successful MEV transactions;
• approximately 716 million State-Invariant Transactions;
• success rate — about 2%.

In other words, for a single trade to turn a profit, the algorithms had to lose hundreds of other trades

---

This is exactly where the MEV economy is changing

Most public reviews evaluate only successful transactions.

After this recalculation, it became clear that a significant portion of speculative strategies on Layer 2 have lost much of their appeal. The reason is not that arbitrage has stopped working, but that the cost of constant competition has risen sharply.

The winner still makes a profit.

But dozens of losers are footing the bill for this victory.

---

Flashbots reaches similar conclusions

Flashbots researchers have also previously highlighted the implications of low-cost Layer 2 transactions.

The lower the transaction fee, the cheaper mass competition among bots becomes.

As a result, the network handles not only actual user transactions but also a huge number of attempts by algorithms to be the first to be included in a block.

Thus, a decrease in the price of natural gas simultaneously increases the volume of competitive MEV spam.

---

Conclusion

The work *There Will Be Spam* is interesting in that it is the first to shed light on the economics of the losers—that part of the market that almost never makes the headlines.

For a single algorithm to succeed, dozens of others must first fail, pay fees, and take up space in the blocks. It is precisely these failed attempts that account for a significant portion of network activity and force us to take a fresh look at the statistics of modern blockchains.

Conclusion by KLYO

Studies like these make me look at blockchains from a completely different perspective...

But who is actually driving this activity?

People?

Or thousands of algorithms that compete against each other around the clock, paying the price of millions of failed attempts for the sake of a single successful trade?

I'm not saying that blockchains are empty. The study doesn't say that.

But after reading it, I was left with one question.

How many of the users on modern blockchains are actually real people, and how much of the activity is generated by machines?