Blog | BLAKE2b MINER

blog / blake2bminer.com /

The front page is the reference. This blog is the record. @BLAKE2bMiner is the feed. Position disclosed: we mine it, solo, through our own node. It started with The producers can leave .

2026

The rentable ceiling

The hash floor post asked what renting hash to attack this chain would cost. This one asks whether an attacker could rent enough hash at all. Today, no. The number to watch is how much of today's hash is rented already.
The hash floor, in 88 seconds

The hash floor post as a short animated explainer. One diagram builds through it: what the chain pays, what an attacker's rent costs, and the floor where they cross. Then the four floors against today's hash, the two windows, and the lock. Drawn by code from the post's numbers, with a synthetic voice.
How a year became 180 days

The coinbase lock went from 45 days, to a one-year hold on the table, to a rolling 180 days in six weeks. The dated record, each line a source and a quote, for a miner who wants to know what rule will apply to next month's coins and where that is being decided.
The hash floor

How much honest hash does this chain need before renting hash to attack it costs more than it takes? A floor in PH/s from today's rental prices and today's coin price, as a range, and the one ratio it turns on. The chain sits at the floor, not above it.
The empty suit

They bought the stage, the ticker, the fund, and most of the hash. A pleb in a spare room runs a node and checks every block against the rule. One of those is empty.
Shorten the window

We replayed the chain's own hashrate under forty versions of bitcoin's difficulty rule, eight window lengths by five caps. A 144-block window with no cap pays a week-long 4x raid 444 coins instead of 5,906, and the miners who stay are back to pace in 5 days instead of 60. The cost, in one table.
Delete the part

Nine ideas for the surge in four weeks, each one fixing what the one before it broke. The ledger of doors, with the number that decided each. A 1945 result that says why the slow difficulty rule is also what makes cheating expensive. Then the test: the cap was a fifth of the surge, the window paid the rest, and bitcoin's rule loses in both directions.
The first day of the 45-day wait

The 100-block grace ended at 973,540 on September 22. In the next 24 hours our node rejected ten blocks, each for spending a coin mined at 973,440 or later; one spend was mined nine times; nobody built on any of them; no coin from the window has moved, and the ledger's spent share went from 65 to 93 percent a day to zero. The sixth retarget landed at 1.353x, the second the cap did not touch. One dated record, from one node.
Upgrading an Umbrel node to Knots 29.4.2

The 45-day coinbase maturity enforces from block 973,440. Moving the official Umbrel Bitcoin Knots app from rc4 to the 29.4.2 final took one command, one click and 22 seconds of downtime. The store trap, the way back, the node's own before and after, and one catch: the release changed the difficulty RPC field and the mempool app stopped.
The one-part fix

The surge paid 26,112 XBT above bitcoin's schedule because difficulty recomputes every 2,016 blocks and moves at most 4x a step. Two posts here scored reward rules that patch that lag. This one replays the chain's real hashrate from our node under a fast difficulty rule instead, zero new parts, and finds it removes 83 percent of the excess on its own. What it leaves, what it costs in timestamps, why the extra-work factor removes almost nothing, and the eighty-year-old field that already ran this argument. An explainer, not a proposal.
Pay for share

A variant of pay by the clock appeared within twelve hours: give every seven days, or every day, a fixed budget of coins and split it equally across the blocks found in it, settled after the epoch closes. Nothing to withhold, no rich or lean blocks. What it would have paid on this chain, from our node; the five points we sent and what happened to them; Zawy's endorsement; where the settlement mechanism breaks, three ways; and the withholding math that corrects our own rule. An explainer, not a proposal.
Pay by the clock

On September 1 the chain paid 5,573 XBT in a day at 31-second blocks, because the subsidy is per block and the retarget could not catch up. Paid by the clock, 3.125 XBT per 600 seconds, that day pays 454 and every day pays about 450. What one rule would have done since activation, computed from our node, what it does to the incentive to add hash, and what it breaks. An explainer, not a proposal.
Who got the coins

Mining Stacker says anyone who did not come out of the first days with two to five hundred coins got rugged by their own people. We asked our node which window that was, who got how many coins in it, and how much of that has moved. Twelve payout addresses of 793 crossed 300 XBT. Two are pools. Of the other ten, four have moved nothing and four have moved everything.
One address, three names

We went looking for AlphaPool's other addresses with our own node. What we found: none. What we found instead: the same address under three coinbase tags, which is why an explorer's pool list reads it four points low, and a pool that pays about 950 addresses at the hour its coins mature.
The mercenary hash problem

One pool took half the blocks of a ten-day profit window that bitcoin's own retarget rule held open, and the Knots room is now picking a coinbase-maturity number to answer it. The problem named, from a five-hundred-year-old chapter; the three harms separated; and one test, from the white paper, that scores every proposal on the table.
What a block was worth before the market opened

A US miner who received XBT before September 1 owes tax on a value no market had set. The IRS rules that apply, quoted from the PDFs; the coin's first eight days of price from the exchange's own API; and one method, applied to a hypothetical reward, with its weak point marked.
The ticker people will type

This week two exchanges relabeled the BLAKE2b coin XBT and a third still calls it BTCB2. We typed XBT into the one app most newcomers already have and wrote down what came back. Why that matters more than whose ticker it is, and which one we use.
Eighty-three inputs

On September 6 the Liquid federation's wallet paid out 4,019 BTC to someone who had minted L-BTC out of nothing. We ran that transaction through our own node on the BLAKE2b chain. It fails, and the reason it fails is the part worth knowing if you have spent old coins since August 30.
The UPS has no timer

The obvious way to wire a UPS to a node is a countdown: on battery, wait N seconds, shut down. Ours has none, on purpose. Why the timer was wrong, the numbers we measured instead of assumed, the setting the hardware would not let us change, and the one thing still unverified.
Your Goldshell BOX has no lock on the door

The SC Box web API answers reads and writes with no password, and the box ships with an open WiFi access point. What we confirmed, what we deliberately did not test, and what to do today.
Goldshell SC Box firmware 2.2.5: eight degrees cooler, same hashrate

Flashing a used SC Box from 2.2.1 to 2.2.5: where the file is, the checksum, the one setting that decides whether it runs quieter or louder, and our before-and-after numbers.
Gresham's Law mined a block

128 coinbase tags in the first 1,818 BLAKE2b blocks, against 18 pools on the chain that kept SHA-256d. What the chart shows, what it cannot, and which law of money the names are betting on.
Two more fours

The first retarget cut difficulty 41%. The next two are capped at 4x each, and the arithmetic says half-minute blocks for a day, then two-and-a-half-minute blocks for three and a half days, then ten. Dated, so it can be checked.
Three blocks on the first night

Our $250 used Goldshell SC Box found three BLAKE2b mainnet blocks in eight hours, about 9.39 BTC of coinbase to our own address. What the odds actually were, why the explorer files us under Unknown, and what a reset would do to all of it.
Where the coins are, and why your wallet cannot see them

Three blocks paid 9.39 BTC to our Ledger address. Ledger Live will not show them, Sparrow would not either, and mempool.guide shows all of it. Nothing is wrong. Here is the picture.
Sealed box to hashing: setting up a used Goldshell SC Box

The hardware half of solo mining BLAKE2b: a $250 used ASIC, a desktop PSU, a jumper switch, and every trap between the eBay box and accepted shares.
Getting an Umbrel node onto BLAKE2b mainnet today

The exact steps that took our Umbrel from a stalled SHA-256d node to solo mining BLAKE2b mainnet: Knots rc4, the wallet, and DATUM, with every trap we hit on the way.
The producers can leave

You don't have to win the argument. You can stop participating on their terms, keep what's yours, and build.

© 2026 BLAKE2b MINER. Quote it, link it. Disclosure & terms

Press Esc or click anywhere to close