Robots built from Bitcoin’s own blocks. Every block keeps the winning number its miner found, and the machines that search for those numbers leave marks in them. A Minitor reads the marks in one block and becomes the robot they describe.
Minitor No. 388,067, running live. Tap its screen to flick through the pages, tap the robot to make it mine, and tap the gear for the maker’s mark.
What a Minitor is
A Minitor is a robot drawn live from one block’s own data. Nothing about it is picked by hand: the block decides its kind, its colours, its grade and every number on its screen.
123456
The robotIts kind comes from the marks in the block’s winning number. There are eight kinds.
The chest screenTap it to flick through eight pages of facts about the block.
The metalTrims, eyes, fans, pens and springs upgrade with the block’s luck, from Iron to Diamond.
The colourThe background shows which halving era the block was mined in.
The details panelGroup, what the block shows, era, time and winning number, at a glance.
The gearTap it to see where Minitors come from.
The winning number
To add a block, a miner has to find a number that makes the block’s hash fall under a target. The block keeps that winning number forever, so anyone can read it.
The winning number is 4 bytes long, about 4.3 billion possibilities. Each mining machine searches that space in its own way. Some only ever use certain values, and some can never produce others. Over hundreds of thousands of blocks those habits show up as patterns you can count. Minitors read four of them, with three more years of no proven pattern in between.
Satoshi’s set
In 2009 and 2010, the last byte of the winning number lands in two ranges far more often than chance: 0 to 9, and 19 to 58.
Sergio Lerner described this pattern in 2013 and linked it to the machine that mined many of Bitcoin’s first blocks, widely thought to be Satoshi’s. In blocks 0 to 36,288, 65% of winning numbers end in those ranges. By chance it would be 20%.
The last byte of each winning number, blocks 0 to 36,288. Each bar is one value from 0 to 255; orange bars are Satoshi’s set. Hover a bar for its count.
A Minitor never claims who mined a block. A 2009–2010 block whose winning number ends in that set simply becomes a Satoshi byte robot.
Counting from zero
Small winning numbers are common in Bitcoin’s first two years.
In 2009 and 2010, 44% of winning numbers sit in the bottom sixteenth of the range. In 2012 to 2014 it is 6.2%, exactly what chance gives. That fits early miners that counted up from zero and started again often, so small numbers came up most. A 2009–2010 block with a small winning number, and no Satoshi byte, becomes a Low number robot.
Machines with empty slots
From mid-2016 to 2019, seven of the 64 values in the top part of the winning number turn up only a third as often, because the most common machine of those years could not make them.
The top six bits of the winning number work like a slot number, from 0 to 63. In 2011 to 2014, every slot is used about equally. In 2015 and early 2016, slots 50 and up are used less often. From mid-2016, slots 57 to 63 drop to about a third of the rest.
Slots used normallySlots that era’s machines used lessDashed line: what chance gives (1.6%)
How often each slot appears, as a share of each era’s blocks. Hover a bar for its exact figure.
The S9’s missing slots
The Antminer S9 mined a large share of blocks from 2016 to 2019. Each of its chips has 114 working cores but numbers them in a space built for 128, so a few slot numbers never come out of an S9. Bitmain’s own S9 code shows this numbering, and Coin Metrics published a study of these machine fingerprints in 2023. The 7,536 blocks from those years that do land in slots 57 to 63 were found by other machines.
A block from mid-2016 to 2019 that an S9 could have made becomes an S9 fit robot. One it could not have made becomes a Gap jumper.
The 2015–16 pattern
The 2015–16 pattern is clear in the blocks, but we found no maker’s code that explains it. So those robots only say what the data shows: an S7 era robot fits the slot pattern of 2015–16 blocks, and a Gap jumper from those years sits in a slot they rarely used.
Scribbles in the version
From 2020, Minitors read a different mark: scribbles in the block’s version field.
BIP 320 sets aside 16 spare bits of the version field that miners can use as extra search space. A block that uses them has scribbles in its version. Almost no blocks did before 2018; in 2019 it was 44%, and in 2025 96%.
Share of each year’s blocks whose version field has scribble bits.See the numbers
Year
Blocks with scribble bits
2015
0.0%
2016
0.1%
2017
0.0%
2018
5.3%
2019
44.1%
2020
68.4%
2021
73.2%
2022
84.3%
2023
92.2%
2024
94.7%
2025
95.9%
From 2020 on, a block with scribble bits becomes a Scribbler and a block without them becomes Plain, so Plain robots get rarer every year.
The eight robots
Each block becomes one of eight robots, sorted by fixed rules from its year and its marks.
Satoshi byte
2009–2010
The winning number ends in Satoshi’s set.
43,818 blocks
Low number
2009–2010
The winning number is in the bottom sixteenth of the range.
34,058 blocks
Static
2009–2014
Matches no fingerprint that can be proven.
258,985 blocks
S7 era
2015 to mid-2016
Fits the slot pattern of 2015–16 blocks.
64,143 blocks
S9 fit
Mid-2016 to 2019
A winning number an S9 can make.
183,155 blocks
Gap jumper
2015 to 2019
In a slot those years’ machines rarely used. From mid-2016, a number an S9 cannot make.
26,532 blocks
Scribbler
2020 on
The version field has scribble bits set.
266,955 blocks
Plain
2020 on
The version field has no scribble bits. Rarer every year.
48,394 blocks
Block counts cover blocks 0 to 926,039, mined up to 1 Dec 2025. Gap jumpers: 18,996 from 2015 to mid-2016 and 7,536 from mid-2016 to 2019.
Luck and grades
Every block’s hash must fall under a target. Most land just under it; a few land far under. That distance is the block’s luck, and luck sets its grade.
Luck counts how many times smaller than the target the hash is, in doublings: level 3 means at least 8 times under, level 10 at least 1,024 times. Each level is half as common as the one before. Across every block, 905 reached level 10 or more; chance predicts 904.
Grade
Luck level
How rare
Blocks so far
Iron
0 to 2
Most blocks
810,501
Copper
3 to 5
1 in 8
101,067
Silver
6 to 8
1 in 64
12,706
Gold
9 to 11
1 in 512
1,535
Platinum
12 to 14
1 in 4,096
211
Diamond
15 and up
1 in 32,768
20
IronCopperSilverGoldPlatinumDiamond
One S9 fit robot drawn at each grade, to show what changes. Real blocks set their own grade.
What changes as the grade goes up
Copper: metal knee bands, wrist cuffs and eye rings.
Silver: eyes that glance around, a glass shimmer, scanning visors, re-rolling counters.
Gold: a glow and twinkling stars, glowing eyes, faster fans, a flipping coin, a gold pen that drips ink.
Platinum: a light sweep over the body, scanning rings, a glowing stylus, glowing springs.
Diamond: sparkles orbiting the robot, gold plating, crystal parts and a double hop.
Only 20 blocks have reached Diamond so far. The luckiest, blocks 368,527 and 756,951, reached level 20: a hash over a million times under its target.
Minitor No. 368,527, a real Diamond.
Halving colours
The background colour shows the halving era the block was mined in, and every future era gets a colour of its own.
Era 12009 to 2012, 50 BTC reward
Era 22012 to 2016, 25 BTC
Era 32016 to 2020, 12.5 BTC
Era 42020 to 2024, 6.25 BTC
Era 5From 2024, 3.125 BTC
The code works out each later era’s colour from its number, so a block mined in 2040 already has its look.
The chest screen
All the facts live on the robot’s chest screen. Tap it to flick through eight pages.
1. MinitorIts group, what the block shows, its grade and era.2. MarkersWhich of the marks this block has.3. Winning numberThe winning number, its last byte and its slots.4. BlockHeight, time, version, reward, and the sats the Minitor holds now.5. Why this groupThe exact rule the block met.6. LuckThe hash replayed, the luck level and grade.7. Upgrade votesWhich network upgrades the block voted for.8. Slot mapIts own animation. For this block: a map of the slots.
These are the pages of Minitor No. 388,067, a block that voted for the BIP65 upgrade in 2015.
How to mint
Bitmap holders mint Minitors on the Ordinal Portal: one per bitmap, as a child of the bitmap.
Open the mint page on the Ordinal Portal.
Connect the UniSat wallet that holds your bitmap.
Pick your bitmap. The page checks it is the real first-is-first bitmap, that it has no Minitor yet, and that it is settled in your wallet.
Check the preview. That is your Minitor, drawn by the inscribed code from your block.
Pick a speed and approve one popup in UniSat.
Your Minitor lands in your wallet once the next block or two is mined, usually 10 to 20 minutes.
What it costs
There is no mint price. You pay the Bitcoin network fee, about 800 sats at 2 sat/vB (it rises and falls with traffic), and 330 sats that stay inside your Minitor. The page shows the exact total before you approve.
The rules
Real bitmaps only. The page checks the bitmap against Bitoshi Blockamoto’s bitmap rules and bop’s LiveOCI, right up to the current block.
One Minitor per bitmap. The first one minted is the real one.
Your bitmap is the parent. It is used in the mint and comes straight back to your wallet, whole, in the same transaction.
It never touches a parcel. A Minitor is a child of your bitmap but not a parcel, so it never uses up or blocks one.
Mint a MinitorYour wallet signs everything. The page never sees your keys.
Fixed forever
The robot code is inscribed once and every Minitor uses it. There is nothing to update and nothing stored anywhere but Bitcoin.
Each Minitor reads its block number from its parent bitmap and its data from the block itself. It shows only facts: it never says who mined a block. The one hard “no” it makes, that an S9 cannot make a winning number, comes from Bitmain’s own S9 code. New blocks keep coming, and each future block’s Minitor is sorted by the same fixed rules.