Feedback from a bitcoin user interview - letter “l” mistaken as numeral 1

[xml]

Feedback from a bitcoin user interview - lowercase letter “l” mistaken as numeral 1 in bitcoin addresses (e.g. bech32).

*Example scenario*
Read bitcoin address from a screen/display/paper. Then type on a keyboard in another device. (Possible context: sharing one-time bitcoin address to another user in order to receive bitcoins).

*Result*
Many times users misread lowercase letter l as numeral 1. They share incorrect address, transaction can't be made and so bitcoin is not received. It's not easy for a user to realize a cause of the problem and recover from the situation. They repeat the same misread every time and get even more confused.

*Question*
Can I ask you to respond whether you faced the same problem in order to evaluate the magnitute of it, please? I believe that it happens in majority of cases, especially that 1 is used in every bech32 address (as a separator at least).

(On a side note, I think that users should decide what encoding to use so let's give them many options, e.g. word-safe base20, bech32, base36, HEX, base58, Radix64, base85, US-ASCII and learn from it).

[1715074952]

1715074952

[1715074952]

1715074952

[1715074952]

1715074952

[Pmalek]

Bitcoin addresses have checksums. It's there so that you can't send Bitcoin to a non-existing address like in the case you mentioned. If you take any valid address and replace one of its characters for a different one, the address becomes invalid and most (good) wallets won't allow you to make a transaction to it. 

Another thing. Who types down an address character by character nowadays anyways? You copy/paste the whole thing or scan the QR code then you check if the two addresses match before finalizing a transaction.

To answer your question, I have never had that happen to me because I don't write down addresses character by character, and you shouldn't either.

[NeuroticFish]

Same here, I've never faced this issue because I usually copy-paste (and check thoroughly I've pasted what I was supposed to, since clipboard malware is a real threat) or scan QR code.
Even more, I have experience and I do write different I, l and 1.

"transaction can't be made and so bitcoin is not received. It's not easy for a user to realize a cause of the problem and recover" <- I tend to disagree with this, the wallet should tell that the address is incorrect. At that point, if the user is well intended and has no other solution, will contact you and ask for a correct address.

[o_e_l_e_o]

Base58, which is used for legacy addresses such as P2PKH and P2SH, does not use capital i "I" or lowercase L "l", so there should be no confusion between these characters and "1".

Bech32, which is used for newer addresses such as P2WPKH, does not use the number "1" (outside of the separator character in bc1) or the letter "i", so there should be no confusion between these characters and lowercase L "l". (Bech32 only uses lowercase letters).

So if the address starts with 1 or 3, then the character you can't read is a 1. If the address starts with bc1, then the character you can't read is a lowercase L "l".

But as has been said above, very few people write or copy addresses by hand, and the checksum present in all types of address will almost always prevent you from sending coins to the wrong address.

[Husires]

According to Bitcoin wiki the probability of mistyped identifier of 26-35 alphanumeric characters that is accepted as being valid is 1 in 2³², that is, approximately 1 in 4.29 billion. https://en.bitcoin.it/wiki/Invoice_address#Addresses_are_often_case_sensitive_and_exact

Imagine with me who will try to type the Bitcoin address manually, let's say choosing the shortest address of about 26 characters? Also, most computers and mobile phones support either QR or NFC or both, or at least you can copy and paste the address.

[dkbit98]

Many times users misread lowercase letter l as numeral 1. They share incorrect address, transaction can't be made and so bitcoin is not received. It's not easy for a user to realize a cause of the problem and recover from the situation. They repeat the same misread every time and get even more confused.

I don't know anyone who is typing complete bitcoin address manually, and I never did something like that.
There is a chance that someone could write address on piece of paper, but it's more likely that letter ''I'' would be replaced with letter ''l'', and not number 1.
This is the reason why I think seed words and better than private keys, because they can be written down easier.
To avoid that mistake simply copy-paste, export QR code image, etc.

Can I ask you to respond whether you faced the same problem in order to evaluate the magnitute of it, please? I believe that it happens in majority of cases, especially that 1 is used in every bech32 address (as a separator at least).

I never heard of such case (until now), that means it doesn't really exists (for me).

[o_e_l_e_o]

According to Bitcoin wiki the probability of mistyped identifier of 26-35 alphanumeric characters that is accepted as being valid is 1 in 232, that is, approximately 1 in 4.29 billion. https://en.bitcoin.it/wiki/Invoice_address#Addresses_are_often_case_sensitive_and_exact

You need to fix the formatting. 1 in 232 and 1 in 2³² are vastly different numbers.

This failure rate of 1 in 2³² is also only true for Base58 addresses. For Bech32 addresses, the checksum is guaranteed to detect up to 4 errors. Beyond that, the failure rate is less than 1 in 2³⁰.

[nc50lc]

*Question*
Can I ask you to respond whether you faced the same problem in order to evaluate the magnitute of it, please? I believe that it happens in majority of cases, especially that 1 is used in every bech32 address (as a separator at least)..

Not happened to me even once.
Number 1 and lowercase l look entirely different in every 'font face' available.
It will only be realistic if it is uppercase I & lowercase l and number 0 & uppercase O but those are already solved by Base58check and Bech32.

The '1' that's always available in legacy '1' and SegWit 'bc1' addresses actually helps.
Think of it, since only 'l' is being mistaken as '1' and not vice versa (who'll mistake 1 as l anyways),
Having '1' at the first few characters of the address will serve as a reference so the user can already tell which is '1' and which are not.

[xml]

Another thing. Who types down an address character by character nowadays anyways? You copy/paste the whole thing or scan the QR code then you check if the two addresses match before finalizing a transaction.

To answer your question, I have never had that happen to me because I don't write down addresses character by character, and you shouldn't either.

Thanks for a response.

Answering your question, I think that HODLers, paper or hardware wallet users and more type down addresses character by character. Also, there is risk that inexperienced user would misread l as 1 when checking if the two addresses match in your example. Lastly, I can't find a reason why an address shouldn't be written down. Can I ask you to elaborate on your suggestion, please?

Same here, I've never faced this issue because I usually copy-paste (and check thoroughly I've pasted what I was supposed to, since clipboard malware is a real threat) or scan QR code.
Even more, I have experience and I do write different I, l and 1.

"transaction can't be made and so bitcoin is not received. It's not easy for a user to realize a cause of the problem and recover" <- I tend to disagree with this, the wallet should tell that the address is incorrect. At that point, if the user is well intended and has no other solution, will contact you and ask for a correct address.

Thanks for a response.

Following-up on the disagreement, I think that the problem exists even if a wallet informs that the address is incorrect. A user may not know what exactly is incorrect and repeat the same mistake every time reading an address.

I don't know anyone who is typing complete bitcoin address manually, and I never did something like that.

I never heard of such case (until now), that means it doesn't really exists (for me).

Thanks for a response.

Not happened to me even once.

Thanks for a response.

[o_e_l_e_o]

Answering your question, I think that HODLers, paper or hardware wallet users and more type down addresses character by character.

I've never done this. I've printed a few addresses, but never by hand. If I'm writing something by hand then it is always a seed phrase, which were partly created to be much more resistant to transcription errors.

Lastly, I can't find a reason why an address shouldn't be written down. Can I ask you to elaborate on your suggestion, please?

For this exact reason. People read things wrong and people have terrible and often illegible writing.

Following-up on the disagreement, I think that the problem exists even if a wallet informs that the address is incorrect. A user may not know what exactly is incorrect and repeat the same mistake every time reading an address.

Bech32 checksums actually include error correction as well as error detection. I'm always hesitant to suggest using this ability, though, as it is invariably only needed by inexperienced users, and when the error correction algorithm suggests a correction, such a user is likely to accept the correction without double checking it and then send their coins off to the wrong address.

Pieter Wuille has a tool which does something in the middle, showing you where the errors are but not showing you the correction in case it is the wrong correction. You can play around with it here: https://bitcoin.sipa.be/bech32/demo/demo.html

[Pmalek]

Answering your question, I think that HODLers, paper or hardware wallet users and more type down addresses character by character.

I am a hodler and hardware wallet user and in all my years in Bitcoin, I don't have a single backup of my addresses. There is no reason for you to have them. You aren't sending Bitcoin using your local post service where you have to write down the address on a piece of paper or receipt. It's done digitally and over the Internet. If I give you the link to this thread (https://bitcointalk.org/index.php?topic=5404332.0) and ask you to save it, you are not going to take a piece of paper and write it down by hand, are you? You will bookmark it, copy and paste it into notepad, or something similar. Your seed is the most important element of your wallet. That's the only thing that should be written down on paper safely and carefully. The seed takes care of the rest. It derives your private and public keys, and your addresses. 

Also, there is risk that inexperienced user would misread l as 1 when checking if the two addresses match in your example.

Another reason why you shouldn't write it down by hand, isn't it?

Lastly, I can't find a reason why an address shouldn't be written down. Can I ask you to elaborate on your suggestion, please?

You shouldn't search for answers why it shouldn't be done, but ask yourself why you should do it? Let's say I have used 100 addresses in the last year. What possible benefit would I have to keep all these addresses written down on paper? The address itself can't spend coins without its seed and private key. Like I mentioned previously, having multiple copies of your seed on paper is the only thing you need to do.

One more question for you:
If you need to send me Bitcoin, how would you prefer that I give you my BTC address? Should I write it down by hand on a piece of paper and send you a photo of me holding the paper in my hand, or do I copy/paste it and send it you via PM, email/Telegram or any other digital way?

[xml]

I've printed a few addresses, but never by hand.

So, eventually, you may have faced the problem of misreading the lowercase l character.

One more question for you:
If you need to send me Bitcoin, how would you prefer that I give you my BTC address? Should I write it down by hand on a piece of paper and send you a photo of me holding the paper in my hand, or do I copy/paste it and send it you via PM, email/Telegram or any other digital way?

If BTC addresses are to be used (written, sent, read) in digital ways only then why bothering about encoding (bech32, base58 etc.) at all? Why would it make any difference whether they are human readable, printable or exclude indistinguishable characters? Similarily, are URLs (re-)typed or read in digital ways (clipboard pasted or copied or scanned) only?

[o_e_l_e_o]

If BTC addresses are to be used (written, sent, read) in digital ways only then why bothering about encoding (bech32, base58 etc.) at all? Why would it make any difference whether they are human readable or exclude indistinguishable characters?

To make them easy to double check. Just because it isn't great to hand write private keys or addresses, doesn't mean you shouldn't be double (or more) checking everything you copy and paste.

Similarily, are URLs (re-)typed or read in digital ways (clipboard pasted or copied) only?

The vast majority of top level URLs that people are typing in are composed of familiar words and not random characters.

[jerry0]

is it possible for someone to copy/paste you a btc address for you to send to... but when you copy an paste btc address... somehow that causes virus or malware or keylogger?  That seem to be one reason to not want to copy/paste when someone send you a btc address.  But that is not possible right?  Because if it was malware or virus or keylogger, then the btc address they post would be more like a link that you click on?


 

[o_e_l_e_o]

is it possible for someone to copy/paste you a btc address for you to send to... but when you copy an paste btc address... somehow that causes virus or malware or keylogger?

It is entirely possible for malware to change an address on your clipboard, so you copy the real address but then paste a malicious address belonging to an attacker. This is known as clipboard malware. This is why you should always double check any address you have pasted against the source to ensure it has copied accurately, as I said above.

Because if it was malware or virus or keylogger, then the btc address they post would be more like a link that you click on?

It would also be possible for a piece of malware to change a clickable bitcoin payment request to a URL which then loads in a browser, sure.

[Pmalek]

<Snip>

Now you are just going to confuse jerry0 further. He is either going to ask what a clickable bitcoin payment request is or he is going to think all Bitcoin addresses are also clickable bitcoin payment requests.

What jerry0 wants to know is basically this: can a random Bitcoin address, such as bc1qreec5wqs7czc2a0julnfchrxn8039gjsww7h94 (I just took one from a blockchain explorer) be infected with some kind of malware in the form it is now? A string of random characters. He has of course asked this in the past as well because that's jerry0 for you. Why ask it once when you can do it every year? So, no jerry, it can't. bc1qreec5wqs7czc2a0julnfchrxn8039gjsww7h94 is not a malware-infected piece of code and you can safely copy and paste it without wearing gloves and a mask.   

[xml]

Let's get back to the subject and the question "whether you faced the same problem [letter “l” mistaken as numeral 1] in order to evaluate the magnitute of it, please?".

[Pmalek]

I think o_e_l_e_o gave you a very good answer already.

If we are talking about SegWit addresses, the character "1" only appears in the beginning as the 3rd character of the string - bc1. The lowercase "i" is not part of the address. If you remember that, you can't confuse those two with "l".
In Legacy addresses, you won't come across a capitalized i  - "I" or a lowercase L - "l". 

However, o_e_l_e_o did make a slight mistake when he said:

(Bech32 only uses lowercase letters).

Bech32 addresses are not case-sensitive, but they can be either in uppercase or lowercase letters. But you can't mix them two. They are either all capital letters or lowercase ones. "bc1q"... is equally valid as "BC1Q...", but you will rarely come across the 2nd variant. That's why it's not really a mistake because most people represent them in lowercase letters. In contrast to Bech32 addresses, Legact and Nested SegWit addresses are case sensitive and its normal to have a mix of uppercase and lowercase letters.   

[nc50lc]

Feedback from a bitcoin user interview - lowercase letter “l” mistaken as numeral 1 in bitcoin addresses (e.g. bech32).

*Example scenario*
Read bitcoin address from a screen/display/paper. Then type on a keyboard in another device. (Possible context: sharing one-time bitcoin address to another user in order to receive bitcoins).

I refuse to believe this scenario.
If it's real, it's either: 'l' and '1' were written in a homogenous way for the majority of users to mistook the characters or it's actually an eye checkup for prescription glasses.

Any source of that 'interview' so we can "evaluate" its magnitude?

[o_e_l_e_o]

Bech32 addresses are not case-sensitive, but they can be either in uppercase or lowercase letters.

This is true, but I'm not aware of any website or service which uses uppercase letters for Bech32 addresses. Lowercase letters are ubiquitous. But still, since Bech32 addresses do not use "1" outside of the separator, and do not use "i" at all, then whether it is "l" or "L" there is no other character it can be confused with.

I refuse to believe this scenario.

It would be incredibly unlikely when reading from a screen, since most fonts quite clearly differentiate between lowercase l and the number 1. It would be very possible when reading from something hand written, since many people write the number 1 simply as a vertical line.