Multi-sig : our talented nephew working at the shoe store
In 2014, an eon ago in bitcoin time), Gavin Andresen declared it the year of multi-signature at the Amsterdam bitcoin conference1.
The idea is to “distribute the signing keys” over a threshold, to require a few signatures instead of one, and your bitcoin suddenly felt safer. The only problem solved is the short term distribution of signing keys. That’s useful for some companies with a specific setup of course. And there it more or less ends in my opinion. At least for now.
People back in 2014 thought this was the security upgrade that would make Bitcoin ready for the “grown-ups”. Since then, along with hardware wallets (the two industries go hand-in-hand), they sell the feeling where multisignature is where you feel “better” and more secure. It’s like when one would star drilling holes in a fireproof door and adding some fancy bicycle locks to it.
It probably feels better to some. More safety much.
Multisignature had that clean technical elegance engineers love.
And hardware wallets gave you something to hold on to, something physical to make that ungraspable blockchain feeling physically present. It’s also nice to play around with these devices. It’s our magic the gathering seasonal card set.
Looking back, multisig feels like that talented nephew who just got hired at the shoe store and starts redesigning the whole inventory on day one. The plan looks good.
Time is what ruins it usually.
Dead sea, dead seed
There is an old copper Dead Sea scroll nr. “3Q15”.
It is a treasure map, literally2.
It describes the location of golden coins and silver, exact spots, cisterns, walls, tunnels. Exact positions described with extreme accuracy to retrieve the valuables.
On paper it looks precise enough to start digging immediately, and some teams actually did just that… to come home empty handed.
Because every useful piece of context around those instructions has since evaporated. The language drifted, some landmarks vanished. The city they used as a reference in the scroll, now sits under sediment layers or got moved by rivers that no longer follow the same course. The bridges and trees are all gone.
Just like software upgrades can shift. The landscape and history itself erased all of the context.
The people who knew the codes and references are long gone as well. The makers of that scroll probably never pictured their careful pointers disappearing under thermodynamics. The universe doesn’t care either, value is irrelevant for a moving riverbed.
That quiet universal decay I’m illustrating here, also hits everything physical and digital around you, given time.
You can delay it. You can’t beat it. You won’t survive time.
Multi-signature runs into that problem rather fast because it has more moving parts, and those parts are too fragile in my opinion.
It starts with the devices used.
Hardware wallets in my own experience last about 2 to 4 years. The flash memory inside is not a stone tablet either. No matter the “military grade” marketing terms their manufacturers throw around.. Charge leaks, bits drift, screens go blank and some usb connectors or batteries fail.
Hardware wallets store seeds in flash or EEPROM, not DRAM. Unlike DRAM, these cells do not need constant power, but the trapped charge still leaks over time, even faster if the device is worn or stored in a warmer place. This retention is measured in years, not infinite.3
The decay is everywhere in different forms: devices, but also coordinator software, developers going broke and going rogue, disappearing code, contributors who leave or “have a great new idea”.
The recovery notes people wrote and didn’t mention one crucial detail in4. All of it starts “rotting” the moment you stop maintaining the machine.
Slow and Furious
The moment you’re in need for quickly replacing one part of such a multisignature setup, you’re also in a race to get your coins out in time (like with a 2 out of 3 multisig with 2 coldcard Mk3 models for example).
Before you know you’d be hopping on a plane to Guatemala to dig up an old hardware wallet under a tree next to a family member’s vacation home.
Good luck winning that race against whoever is exploiting a vulnerability against your coins at that particular moment. The calculating power and speed of setting up such an attacks only goes up over time. While your moving parts’ collision with universal dynamics goes up.
What many people tend to forget about a multi-sig setup is that you actually need to babysit it, replace some “legs” of it when something happens, or re-check every so often to make sure it still performs well. With every check, you expose or risk some more with your setup.
The real backups are usually on paper in some vault, old plastic map or behind the living room painting of a weathered fisherman at your mom’s.
So why not rely on that in the first place?
Your multi-signature setup is an elaborate marketing gimmick to sell more hardware wallets and services maintaining the illusion.
We as bitcoiners have already an extremely good, safe, option : a 256 bit key. It’s secure enough for the foreseeable future and we’ll tackle the quantum problem soon enough I’m sure we’ll have a real security upgrade,... not a fake one.
The 2014 era of multisig never really came to be for the individual user, however. The adoption happened at the corporate level (as explained above, they have real use for it in the short-to-medium term and can babysit the setup daily. Sometime that’s their entire job!).
The everyday user, however, rightfully saw the clunky (or downright unusable) user interfaces of many tools and the incompatibility between different hardware brands and software.
And yes, the user interface of these tools has improved a lot in the last couple of years. A 2016 multisig setup and a 2026 multisig setup are vastly different. However, most of it depends on coordinator software (maintained by a handful of developers) and a descriptor you really need to keep safe. Future compatibility, however, is not guaranteed enough to my liking, whereas a classic Bitcoin wallet (a 24-word one, for example) is.
It is like that talented nephew trying to be a business consultant and overhauling the whole operation. He’s just not there yet.
The Multi-vendor El Dorado
In 2026 this same idea of “old” multisig wears a new label. They call it multi-vendor so you will buy more different brands of these glorified USB sticks.
The backup is still a twelve or twenty-four word list hidden somewhere. The value lives in that 256 bit secret, not in a plastic brick with a marketing team. Your safety was never in the device. It sits in the backup. More brands means more apps, more coordinators, more things that fail, get discontinued, or ship a fatal bug.
New models arrive like seasonal fashion and you are told to upgrade your security. It starts to feel like a subscription.
I wonder how many people spent more on these Hardware wallets than on the bitcoin they were trying to lock down on them.
Companies still invent a role between you and that key however. Recovery services, cloud backups, social circles, inheritance management, multi-party setups that need their servers or hardware to stay alive, or even need their guidance for just “feeling” safe.
Take something close to perfect and charge rent on it. That’s their business model (how fiat).
Bitkey is the latest act in this carrousel of decay5.
The app is user friendly. The recovery still leans on a backup you might not fully control over a decade. Hardware breaks. Screens go dark. Memory decays. Software dies. Over ten years a large portion of users can get locked out.
That is universal entropy.
Another example from history:
The “Case wallet” from Cryptolabs promised the same thing and was described as an “insanely secure” multi-sig wallet.
Back in 2014/15. I know a proud owner of such a device and he had less bitcoin on it than the device cost at the time (of course, paid in bitcoin, so the loss of value is immense).
It had a fingerprint lock, cloud backup 2 out of 3, and all of that… now 11 years later: none of the 3 are accessible (the device since died).
It’s all polished at launch, then the software ages, the founders starting their next project… and the devices fails or vulnerabilities cause hackers to go on a rampage.
If you didn’t move your coins in time, one day you woke up being locked out of that fantastic 2 out of 3 keys multi-signature “insane” security theater. Some of that friend’s coins are still there. Hanging around in “multisignature purgatory”.
It won’t be the last.
History tends to repeat itself, mind you. Today we’re seeing the return of such setups, where the shiny new objects lure you into the next purgatory of decay.
Bitcoin was not built to run these smart contract style scripts for ordinary people. MuSig2 is clever though. It can look like a normal on-chain address while still needing multiple parties underneath. In the end, we’ll end up with an on-chain address that looks like being part of a normal bitcoin wallet, and acts like it, and… actually has more complexity behind it than necessary. How neat.
Keep experimenting.
But I don’t recommend storing your life savings in a distributed decay pool.
Leave the choreography to an exchange finance desk with staff and a daily reason to keep it alive. An individual is better off with acid-free paper (or metal plate) and twenty-four words + passphrase.
Holding two private keys in sucha setup is useless without the exact descriptor: script type, paths, every xpub. Multi-signature does not remove the single point of failure. It moves it onto metadata most people never write down cleanly. And if they do, your next of kin will probably be joining the multisignature purgatory heap of coins they can look at, but never move.
You can copy twenty-four words and put the copies in a few boring places. Paper and a pair of scissors cost about two euros and never ask for a firmware update. Satoshi’s coins are still single signatures after seventeen years and surviving every attacker on earth. So are most “unsafe” paper wallets made back during the pre-2014 era.
This coldcard vulnerability and the renewed interest in security is, of course, the perfect excuse to sell people more hardware. Most buyers do not actually need it, and plenty would be better off never dipping a toe into the multisig setup.
The patient attacker looms
Ledger has moved more than eight million units since 2014. Trezor has sold over two million since 2013. Together that is roughly ten million devices across a decade, and their 2025 revenues suggest the market is now shifting somewhere around a million units annually. I don’t know, but I guess many of these units are sold to the same people. Would be super interesting to see if they actually are the equivalent of Magic the Gathering card sets that sell to the same people every season.
That number of glorified USB sticks sold, is a massive attack surface. Anyone knows 99% sure these things will be used for storing value.
A so-called “patient attacker” has all the room they need to seed vulnerabilities that are difficult or even impossible to detect.
A million units a year means a million chances to slip something into silicon, firmware, or the assembly line before the device ever reaches your door.
And we have not even talked about the process that turns raw, biased entropy from the chip’s physical sensors into a fair random number. The whole thing sits inside military grade secure chipsets, and it is all under NDA with the manufacturers. So the black box effect is alive and well, even at the most fundamental level.
If a large-scale attacker ever hits, the person everyone calls dumb, who wrote their seed on paper and stuck it to the fridge with a magnet will still be safer than most owners of multi device carousels.
I truly hope I am wrong here, but I think it is inevitable that something like that will eventually strike the supply chain of a major hardware wallet brand. The Coldcard exploit (July 2026) already opened a lot of eyes.
For an exchange desk with staff, multisig is a tool that still has its place (although I hope they have better systems internally), for the individual it’s more and more of a CaaS (Complexity as a Service).
But the salespeople and their influencers jumped in with both feet, pushing even more units under the multi vendor premise. That only expands the attack surface for a patient attacker even further. Maybe the next failure will not be entropy at all. Maybe it will be multisig thresholds, witness data formation, pre-seed flaws in the chipsets, strange rejection sampling. Who knows.
If 2014 was the year of multi-signature, why are we twelve years later still arguing whether normal people should use it?
Complexity in itself is not security. You can’t “triple” your bitcoin security (like some people claim) by adding more moving parts that form just distributed decay.
Complexity is a cost. And for some even a way to sell subscriptions, their security consultancy and their shiny things they absolutely need to keep selling to the same people over and over again.
Simplicity on the other hand, has never had a marketing budget, and that is exactly why it keeps winning in most operational setups. The math stands, no matter if you secure a thousand sats, or a thousand bitcoin.
The universe is totally indifferent to how much value you put in there.
The rule of thumb can be: “Your multi-signature hates you”.
While a normal bitcoin wallet with good entropy will fail alongside bitcoin itself, multi-vendor, multi-signature scripts can fail, independently from bitcoin AND alongside it.
What should you do instead?
In short: get a good entropy source and stick with the system to create a seed phrase.
Do this offline. The checksum calculation happens on most of these tools.
Since I make such tools myself, I’ll mention them here as a first place to look at how this works. Of course you’re free to explore any equivalent or system that gives you good fair entropy. I encourage you to go out and find other tools and systems.
10 and 20-sided Dice:
https://allesvoorbitcoin.be/bip39tool/bip39_guide.html
52 playing cards:
https://allesvoorbitcoin.be/21tools/1deck/
6-sided dice (by Rushabh Parmar):
https://github.com/RushabhParmar12/dice-to-seed
Checksum (offline) calculator :
https://allesvoorbitcoin.be/bip39tool/
Write down twenty-four words.
Make a copy in a safe place. Never photograph it or put it on a copy machine, never put online, neither in the cloud.
Put the original at a really secure place and the copy in a boring envelope in a drawer. That is the whole setup. No USB connectors, no rechargeable batteries, no app that will stop working in three years because the startup pivoted to something else.
And most of all: shut up about it. No one knows your keys are there, keep it that way.
The math does not care.
Pick your poison, we’re all mired in decay one way or another, no need to accelerate it by putting your bitcoin into more decaying objects than necessary.
Don’t be the next copper scroll (3Q15) locked value.
“Listen up, maggots. You are not special. You are not a beautiful or unique snowflake. You’re the same decaying organic matter as everything else.”
[ Tyler Durden, Fight Club6 ]
---
footnotes;
1
Gavin Andresen calling 2014 the year of multi-signature Quote to use: “This is the year of the multi-signature wallet.” He said it in the State of Bitcoin address at Bitcoin 2014 in Amsterdam. https://www.coindesk.com/markets/2014/12/29/how-2014-became-the-year-of-multisig
2
3Q15, the Copper Scroll Cave 3 at Qumran, 1952. Official label 3Q15. A copper inventory of about 64 treasure spots, not a prayer text. The map is precise on paper and almost useless once the landmarks and living context are gone. That is the decay analogy. https://www.encyclopedia.com/religion/encyclopedias-almanacs-transcripts-and-maps/copper-scroll https://www.facsimiles.com/facsimiles/dead-sea-copper-scroll https://www.historyhit.com/the-copper-scroll-decoding-historys-most-elusive-treasure-map/
3
Decaying NAND flash:
https://rossmanngroup.com/technical-reference/how-nand-flash-cells-store-data
https://www.sciencedirect.com/science/article/abs/pii/S002627142400252X
4
STACK TO THE FUTURE
@stack2thefuture
As promised, here's the full story of how I almost lost half my bitcoin stack with an incorrect PIN code, a broken Ledger, and an incorrectly recorded seed phrase - and managed to recover the funds with the help of Kimi K3. https://t.co/eWq3Hz3xS5
11:48 PM · Aug 22, 2026 · 25.5K Views
26 Replies · 17 Reposts · 147 Likes
5
Bitkey Block’s 2-of-3 setup: phone key, hardware key, server key. No user-facing seed phrase. Recovery leans on cloud backup, delay-and-notify, contacts, and Block staying in the loop. https://bitkey.world/learning-hub/how-bitkey-works Recovery whitepaper: https://bitkey.build/content/files/2023/11/Bitkey-Recovery-Features.pdf https://support.bitkey.world/hc/en-us/articles/24395098961940-What-are-Recovery-Contacts-and-how-do-they-help-keep-my-Bitkey-safe
6
Fight Club (book / movie)
Book (Chapter 17):
“You are not special. You are not a beautiful and unique snowflake. You are the same decaying organic matter as everyone else, and we are all part of the same compost pile.”
Film (1999):
“Listen up, maggots. You are not special. You are not a beautiful or unique snowflake. You’re the same decaying organic matter as everything else. We’re all part of the same compost heap. We’re the all-singing, all-dancing crap of the world.”