Bitmain Embroiled in a New Scandal

The Chinese bitcoin miner manufacturer allegedly made some of its machines generate more empty blocks.

Bitmain caught in the act

The ocean pool has faced numerous criticisms lately due to an abnormally high number of empty blocks. Three out of thirty blocks (10%!) found since its launch contained no transactions.

Before going any further, let’s explain what a pool is and why they sometimes create empty transaction blocks.

A pool is a kind of cooperative that prepares the blocks on which the miners work. They receive the rewards, which they distribute to miners in proportion to the computing power provided. This system allows miners to smooth their incomes.

Foundry, Antpool, F2pool, Viabtc, and Binance pool are the main pools. They collectively found 45,707 blocks in 2023. These five pools found more than 84% of the blocks. That’s a lot for a decentralized network…

Knowing that the F2Pool pool probably belongs to the same controlled nebula of Bitmain, which also includes Antpool…

Shares of #bitcoin📷 blocks found by pools in 2023. pic.twitter.com/YF3J4ZzW5B

— Nicolas Teterel (@NTeterel) June 3, 2024

The centralization of the network is even worse than it appears if you look closely. The founder of the Ocean pool stated this weekend that many small pools are actually just fronts for Bitmain. This allegation is based on three things:

1. Some small pools clearly use blocks actually prepared by Antpool.
2. They share the same custodian for the rewards earned at each block (Cobo).
3. An agreement exists to bail out pools that are unlucky (our article on this: Bitcoin – A cartel of pools unveiled).

For the founder of the Ocean pool, “Bitmain is not just behind Antpool; this company represents more than 50% of the blocks and manufactures over 90% of the machines used to mine bitcoins.”

Empty blocks

Regarding empty blocks, they have always existed. There were 147 in 2023, representing 0.27% of all mined blocks.

The existence of empty blocks is due to the fact that a pool has two choices when it finds a block:

• Send a block containing the following transactions to its miners.
• First, send an empty block containing only the previous block’s hash. The advantage is that this block arrives faster. The previous block’s hash is the only absolutely necessary data for a miner to start working.

Explanation:

The data miners hash (run through the SHA-256 algorithm) is in the header of a block. This header contains six pieces of information: Version number / Previous block hash / Merkle tree root hash / Time stamp / Target hash value / Nonce.

The nonce (short for “number only used once”) is a number miners vary frantically until they find a valid block (i.e., a hash of this header that starts with a certain number of zeros).

All this to say that quickly receiving the previous block’s hash allows miners to hash for a few extra seconds. And the fact is that blocks are sometimes found in this time frame, resulting in empty blocks.

Yes, some blocks are found just seconds apart. Blocks are not found exactly every 10 minutes. It’s not uncommon for an entire hour to pass without a block being found.

Fishy behavior

We are getting to the heart of the problem. Given the odds, it is not normal for 10% of the blocks found by a pool to be empty.

The apparent bad luck of Ocean has led to investigations concluding that some miners (antminers) take way too long to replace the empty block with the filled block containing transactions.

But rather than a manufacturing defect, it seems that Bitmain has actually deliberately hindered its competitors. Here’s the explanation from @GrassFedBitcoin, the founder of Ocean:

“Why aren’t empty blocks replaced immediately after receiving blocks containing transactions?

Because Bitmain’s antminers are poor. What we didn’t know is that they were intentionally poor. The antminers continue to work on empty blocks – sometimes for more than 60 seconds – despite having long received the filled block with transactions.

We initially thought the problem was with the design of the ASIC itself (the electronic chip) and thus poor engineering by Bitmain. But it turns out we were wrong; this delay comes from the antminers’ firmware.

We know this because they just released a fix (for the S19 K Pro and XP models). But what are the odds that this fix gets released right after Ocean started publicly raising questions? It’s clear they had a fix and used it for their own antminers.”

Wow just figured out Bitmain's latest fuckery in real time.

I'll give you the back story…

Once upon a time Bitmain was grinding merkle roots instead of nonces aka "covert ASICBOOST".

This gave them an advantage over other miners who didn't know this was possible. One company…

— Mechanic #FixTheFilters #300kb (@GrassFedBitcoin) May 30, 2024

A serious threat to pool centralization

In summary, an antminer is much more likely to find an empty block because it spends more time working on empty blocks. Which has led some to say that the manufacturer Bitmain sells machines designed to be less efficient than those it uses for its own operations.

That’s why MicroBT’s whatsminers and other manufacturers’ ASICs don’t find as many empty blocks…

Furthermore, it goes beyond empty blocks:

“It’s not just the empty blocks that take an eternity to be replaced. This is actually the case for all blocks. In fact, pools constantly update the transactions inside the block as more lucrative transactions come in offering higher fees.

Imagine a super lucrative transaction appears. All pools will immediately send a new block to the hundreds of thousands of miners who (because of Bitmain’s firmware) will ignore it for a while, except the miners working with Antpool [which belongs to Bitmain’s nebula].

It’s not hard to understand Bitmain’s motivations, especially since as time goes on, transaction fees naturally become an increasingly important part of miners’ revenues.”

So, after the”Antbleed” and ASICBoost scandals, Bitmain is once again at the center of criticism. Antbleed refers to the famous back door that allowed any antminer in the world to be remotely shut down. Bitmain defended itself, claiming it was just a way to stop stolen ASICs. Except the code was immediately fixed once discovered…

Another well-known case involves the “ASICBoost” technology. This offers an efficiency gain by reducing the number of calculations needed to generate hashes. This technique saves energy and enables faster hashing.

Let’s conclude by noting that the latest generation of MicroBT miners are already outperforming Bitmain’s. Our article on this topic.

Also, don’t miss our article on pool centralization: Bitcoin and the pool threat.