Topic: Brainwallets re-checked (Read 200 times)

Yesterday I have found that project: https://github.com/dwyl/quotes and I launched search on their quotes.json, but no result. Maybe indeed it is an idea for future (just after your post I found https://github.com/JamesFT/Database-Quotes-JSON/blob/master/quotes.json) but as I said - each quote should be launched in several ways and maybe used as a "root" for other quote (phrase + number? phrase + special characters?).
I am not sure if I want to go that direction, especially that number of potential hits is unknown (maybe there is no single brain wallet founded address anymore?). Of course it is possible that sha256 of random text produces key to existing address, but I exclude possibility of such a luck.

Rather than extract and process it manually from million of books, you could scrape or use API on website which list most popular sentence. Few random example,
1. Quotes, https://www.goodreads.com/quotes
2. Idioms, https://www.theidioms.com/list/
3. Phrase, https://www.phrases.com/top100.php

Probably you are right, but I have a feeling that each of phrase should be modified - internally (remove dots, change to lower/uppercase etc) or externally (add suffix and/or prefix). It complicates things a lot, because if you know that someone used phrase "To be, or not to be", you may play, change text, append year of birth of his seventh son etc.
But if you extract millions of phrases to process (from quotes from books or movies to slogans from commercials)... too much work in my opinion.


;-) Fine. I have missed your post, interesting idea!

It's me.   

Nice work.

Why dn't you try with book words at a future time, instead of with passphrases? You could harvest the XML of an epub book - the text is machine-readable - maybe you could collect them from places such as Gutenberg or archive.org. Then you can uniquify the words using uniq or similar and possibly get greater results.

My gut feeling is that people use as passwords/seeds words of cultural or historical origin such as those found in newspapers, magazines, and other print media.

I have added the results of tests agains 3... and bc1... addresses. As expected, not so many results - 35 phrases, 36 addresses ("secret" was used for both addresses).
But, guys, who of you have used address bc1qc0kwr8clxgj63dp77sr4ql9zhexq2a6aw85ltw from the phase "bitcointalk" ?

Hello

Inspired by recent discussions about brainwallets (https://bitcointalksearch.org/topic/checking-brainwallet-5396293) and (https://bitcointalksearch.org/topic/m.60015668) I decided to repeat the famous excercise of retrieving 18k brainwallets. It was an interesting experience to build a GPU program for that purpose.
I have used list of used addresses for 8th of May (http://alladdresses.loyce.club/all_Bitcoin_addresses_ever_used_sorted.txt.gz) and database of 15462473182 passphrases "all_in_one_p" (https://weakpass.com/all-in-one). I was also considering using extended list of phrases (the one which contains unprintable/special characters) but finally I decided that maybe I will do it later. Because of number of target addresses and memory limitations I had to launch program several times, using each time 30mln addresses for a bloom filter - larger database would make the whole process 3 times longer and I did not expect very different results.
For now, results are only for legacy addresses, today or tomorrow I will launch the same search but for p2sh and bech32 addresses. I am not sure if we may expect any significant number of results.
Results so far: 18995 addresses. I did not check each of them, but from what I observed, many of them were emptied in 2016.
Address 12AKRNHpFhDSBDD9rSn74VAzZSL3774PxQ from phrase "1" is still alive and probably used by bots.
I have uploaded the results there: https://github.com/PawelGorny/BrainwalletsPhrases
If you see strange/suspicious phrase in "legacy_words.txt", try to find it in the file "legacy_results.txt", to see if it was not a address or privkey incorrectly extracted. Unfortunately, for some compresses addresses I have lost the phrase (I had a bug during saving to file, around 20% of compressed addresses has no phrase) - but WIF is there.