Satoshi didn’t leave much behind when he decided to leave the scene for good back in April, 2011. But, he did leave enough for us to conduct a thorough research into his whereabouts when he was working on Bitcoin.
To conduct this research, we gathered data from the following:
- Satoshi’s Bitcointalk account (539 available posts)
- His 34 emails on the cryptography and Bitcoin mailing lists
- His 169 commits on SourceForge
- The metadata from Bitcoin whitepaper versions from 2008 and 2009
- The Genesis block
- Various Wayback Machine archives
The data-driven part of the research focuses on timestamps from Satoshi’s Bitcointalk posts, SourceForge commits, and emails, which represent a total of 742 activity instances from 206 days (not consecutive). The timestamp data starts from October 31, 2008, when he first announced Bitcoin on the cryptography mailing list, and ends on December 13, 2010, when he sent his last email that is known to be UTC timestamped.
Using that data we compiled scatter charts in different suspect time zones to see when he was active and when he was not. We then used other data we gathered to further confirm the most likely location he called home.
Common suspect locations are the UK (GMT), US Eastern (EST), US Pacific (PST), Japan (JST), and Australia (AEST). The last two were easy to debunk, but the first three prospects needed further examination.
Satoshi has posted a total of 539 times on Bitcointalk in the period he was active on the forum – from November 19, 2009, when his account was created, to December 13, 2010, the last time he was seen on the website. Those 539 posts all have a UTC timestamp that specifies when they were published. We know they are in UTC format because accessing the website from different locations shows the same date and time on posts. Since all posts are in UTC format, we can scrape them and plot charts for different time zones. We checked all the three main suspect locations.
Important note: The above chart can easily confuse people who might look at the empty space in the middle and conclude that Satoshi went to bed at about 6 AM and slept until 1 PM in this time zone. However, the instances where his last activity for the day is between 6 AM and 7 AM, for example, are just about 2% of all the dates for which we have activity data. This means that he didn’t usually perform activities until the very early hours of the morning (specific to this time zone), but did so only on a few occasions. In fact, examining the data closely shows that the bulk of his last activity for a day in this time zone is in the range between 1 AM and 3 AM, representing 20% of all days. To counter possible misapprehension, we also plot a last activity median with a range of 6 hours and an arbitrary starting point based on the available data and its representation in each time zone we test.
On first glance, all three locations seem plausible. In London, he is a night owl, working until the early hours of the morning and sleeping until noon. On the east coast, he works all the way up until the end of the day and then sleeps until early morning. On the west coast, he is an early bird, going to bed early in the night, but also waking up very early. Looking at this data alone, we cannot determine, beyond reasonable doubt, which time zone Satoshi lived in.
In the early stages of bitcoin, Satoshi and the few developers that joined the project off the bat were probably working either without SVN or used a local repository. On August 30, 2009, however, the project was uploaded to SourceForge. All commits can still be viewed here. A total of 169 commits were made by Satoshi himself, or as his username was on SourceForge, “s_nakamoto”. Timestamps for each commit still exist in UTC format (proof timestamps are in UTC). Similar to what we did with the Bitcointalk posts, we plot scatter charts and see when Satoshi was active on SourceForge in the three suspect time zones.
The commit charts are consistent with the Bitcointalk post charts. While that is good news, and expected, it doesn’t point a finger to one of the three suspect locations. Still all three are plausible at this point.
After announcing his project on the cryptography mailing list on October 31, 2008, Satoshi continued to communicate with members of that mailing list who became interested in Bitcoin and wanted to know more about it. In a series of 34 emails on the cryptography mailing list and the Bitcoin mailing list, which Satoshi created himself, he discussed the project’s features in length, with a lot of people focusing on the double-spending problem and how Bitcoin solved it. The last email from this batch is from December 13, 2010.
More emails from Satoshi to other people are available here, but we couldn’t verify that the timestamps in those are using the UTC format. This includes his emails to Mike Hearn, Wei Dai, Lazlo Hanecz, Jon Matonis, and Dustin Trammel. So for these charts, we use only the data from the cryptography and Bitcoin mailing lists.
Again, only a confirmation of the pattern we have already established. The email charts still don’t tip the scales in favor of any of the three suspect locations. Now let’s combine all timestamps and see what we get.
Merging all the 539 Bitcointalk post timestamps with the 169 commit timestamps and the 34 email timestamps gives us an even better picture of Satoshi’s activity pattern. In this roundup, the first activity is from October 31, 2008, when he announced Bitcoin on the cryptography mailing list. The last activity is from December 13, when he announced that version 0.3.19 was released.
Anyone else notice the big “S” in the Europe/London and US/Eastern time zones? Anyway, plotting all the timestamp data into a single chart still doesn’t announce a winner location. It can still be any of the three prospects.
By the way, this is a good point to quickly exclude Japan and Australia.
Obviously, Japan is unlikely. While this chart by itself significantly lowers the plausibility of Satoshi calling Tokyo home while he was working on Bitcoin, it is the rest of the facts we uncovered from our research that completely remove Japan as a possible location. We will get to that soon.
Unless Satoshi is a vampire, this one is not even remotely possible and you don’t need any more data to prove that. The next time you are having trouble convincing someone that Craig Wright isn’t Satoshi Nakamoto, feel free to send them the chart above.
In summary, Satoshi’s activity on Bitcointalk and SourceForge, and his emails, all set a clear pattern. Using that pattern we can make reasonable guesses as to where the mysterious inventor of Bitcoin resided while working on the project. The patterns laid out by all the charts above clearly confirm the three most probable time zones – Greenwich Mean Time (GMT), Eastern Time (ET), and Pacific Time (PT).
“Chancellor on brink of second bailout for banks”
Things really start to point towards the GMT time zone, and London in particular, when we look at the Bitcoin genesis block and the embedded message therein:
The Times 03/Jan/2009 Chancellor on brink of second bailout for banks
The key detail here is the title, “Chancellor on brink of second bailout for banks”. This title appeared on The Times (Times of London) newspaper issue on January 3, 2009:
Here’s the problem with Satoshi including this title in the Genesis block if he was living anywhere in the US. This exact issue of The Times did not circulate in the United States. The New York Times announced on May 26, 2006, that the Times of London were planning on distributing a daily US edition of their newspaper in the New York and Washington areas as of June 6, 2006. Setting aside the limited geographical availability of the US edition of the Times of London, it seems that it differs by a lot from the UK version, as one dedicated Times of London reader living in the US found out:
US edition is 64pp and 96pp on Saturdays. It contains the best of The Times – including content from Times2 and the Saturday sections, especially the magazine. Having fewer pages obviously means that we have to be selective! The US edition Page 1 has at least 10 international (not UK) summary/ cross-reference stories (words and pictures). It’s heavy on Business News – there are a lot of summaries on Page 2 under the headline Global Business Briefing. Unlike the UK edition, the news pages open with World News, which runs all the way to the Comment/Opinion editorial section (which is identical to the UK section). Times Britain starts after Oped. Then you get Features – which is the best of Times 2 (or on Saturdays, Saturday Extra). Then Business, then Register, which is just Obituaries. Then a full page of Puzzles and Games (not in UK edition). Then The Game and Sport. There are some exclusive features (eg Michelle Henery’s New Yorker in London column). All our US correspondents and commentators such as Gerry Baker are well displayed – and the fact that the paper is smaller than the UK edition means that you get natural feel of a bias towards American news. It is very much aimed at serious, cosmopolitan business folk in New York and the tri-state commuting area. The Sunday Times would not contain all of the colour magazines (Culture, Style & ST Magazine) you would receive Culture International instead.
This would make it extremely unlikely that front page of the US edition of The Times for January 3, 2009 was formatted in the same way as the UK edition. In fact, judging by the information provided by the above explanation, it is likely that the front page of the US edition for that day contained multiple titles of international news. Maybe the title in question was there, maybe not. I couldn’t find a screenshot of the US edition of The Times for January 3, 2009 anywhere. If you happen to have it, please send it to us.
That being said, who cares about newspapers. In 2009, the internet was already developed enough for media outlets to have websites. Satoshi could have simply used The Times’ website, right? Not exactly. While The Times did have a website already established (timesonline.co.uk at the time), the title of the article in question was slightly different:
Chancellor Alistair Darling on brink of second bailout for banks
It included the name of the Chancellor. This makes it highly unlikely that Satoshi read the article on The Times’ website, because if he did, he would have probably included that exact title in the Genesis block, as opposed to the one from the front cover of the newspaper edition.
At this point, the scales tip heavily in favor of the GMT time zone, and this is without mentioning all the clues that point to Satoshi being British, like his spelling of endings with -ise instead of -ize and -our instead of -or, e.g., analyse, organise, neighbour, colour, etc. Then there’s his use of the word bloody. More specifically, we not only think he is British, but that he also lived in London while working on Bitcoin. Again, the argument is The Times article.
Data from the National Readership Survey from 2008 shows the demographic profile of The Times for that year, where 770 thousand (43%) out of the 1.8 million readership base of the print edition was from London. The rest was split between different areas in Great Britain. This places, almost, every 1 out of 2 people reading The Times in London. In conjunction with all other data we have presented so far, and coupled with London being the largest financial hub in the world, this data makes it highly probable that Satoshi lived in London in the period between 2008 and 2010.
A theory will not be good if we didn’t attack it from all angles. The timestamp data we presented is not up for debate. That is what Satoshi’s activity pattern looked like. But there’s a few things that we need to consider with regards to the Genesis block message.
PDF files contain metadata. When Satoshi wrote the first version of the Bitcoin whitepaper and saved it on his computer, it included a timestamp. More importantly, it included the time zone of the computer. The metadata reads:
This translates to October 3, 2008 at 13:49:58 UTC-7. Daylight saving time (DST) starts on the second Sunday of March and lasts until the first Sunday of November in the Pacific time zone. Thus, the date from the first draft of the Bitcoin whitepaper is from the US/Pacific time zone, which is typically UTC-8, but becomes UTC-7 while DST applies.
In the whitepaper version currently available on the Bitcoin website, the metadata reads:
This translates to March 24, 2009 at 11:33:15 UTC-6. This date also falls into the DST period, but is obviously not Pacific time. It is Mountain Time (MT). States which fall into this time zone include New Mexico, Arizona, and Utah.
There’s a few explanations regarding this metadata. The most obvious one is that Satoshi used virtual machines for some of his activity and set the time zone to whatever. Another possibility is he manually manipulated the time zone of his computer when working on these files.
Then there’s the possibility that Satoshi didn’t create the PDFs himself. He might have written the paper in a Word document and then sent it to someone for a final edit. This might explain the different time zones of the two versions of the paper since he might have sent the second version to a different person.
All that being said, if Satoshi did create those PDF files, and since we know that he was very meticulous when it came to privacy, it is highly unlikely that he forgot to clear/modify the metadata therein. Therefore it is reasonable to assume that he didn’t think that there’s any metadata in the PDFs that would expose him in any way, making the time zone data inside irrelevant.
The Times Digital Access
As we mentioned earlier, Satoshi probably didn’t include the “Chancellor on brink of second bailout for banks” title in the Genesis block after seeing it on The Times’ website since the title of the article there was different from the one in the newspaper. However, web archives show that there was a page on The Times’ website where screenshots of every day’s newspaper cover were uploaded.
Sadly, there is no Wayback Machine snapshot for this page from January 3, 2009, but it is very likely that a screenshot of the cover of the newspaper for that day was present. Moreover, other publications published daily summaries of headline news from newspapers from all over the world. This includes Reuters, which included The Times’s “Chancellor on brink of second bailout for banks” in a comment article on January 3, 2009.
There was also an E-Paper subscription option offered by The Times where subscribers could pay to get digital version of the publication’s newspaper, as is. All of this shows that there were many ways Satoshi could have seen the title he included in the Genesis block, even if he lived in the United States.
However, why would a person go to The Times’ newspapers page or use their E-Paper subscription option to look at the covers of their newspapers when they can simply open the homepage of the website and read the news there? Furthermore, for anyone who has read any of Satoshi’s emails and forum posts, it is apparent that he knew he was onto something exceptional. He tried not to create much hype, but it was obvious he thought Bitcoin could be big. This possibly made the Genesis block an important historical event for Satoshi, such that he decided to include a message in it that will forever stand the test of time. Naturally, this would mean that he would have looked for something thematic, special, memorable. An experience like that can hardly be achieved by reading a roundup article containing 20 headlines or listening to a morning show host reading newspaper headlines.
But maybe Nakamoto heard about the article a few days later and decided it was perfect for a message in the Genesis block. The first Bitcoin block after the Genesis block was mined on January 9, 2009. In this scenario, he learns of the article on January 7, for example, re-writes the code for the Genesis block on that date, and then mines the first block on the next day, creating the first ever BTC transaction between himself and Hal Finney. For those who don’t know, the Genesis block is not mined. It is several lines of text within the Bitcoin source code. Here it is from version 0.1.0 of the software (the first public release by Nakamoto on January 8, 2009):
char* pszTimestamp = "The Times 03/Jan/2009 Chancellor on brink of second bailout for banks";
txNew.vin.scriptSig = CScript() << 486604799 << CBigNum(4) << vector((unsigned char)pszTimestamp, (unsigned char)pszTimestamp + strlen(pszTimestamp));
txNew.vout.nValue = 50 * COIN;
txNew.vout.scriptPubKey = CScript() << CBigNum("0x5F1DF16B2B704C8A578D0BBAF74D385CDE12C11EE50455F3C438EF4C3FBCF649B6DE611FEAE06279A60939E028A8D65C10B73071A6F16719274855FEB0FD8A6704") << OP_CHECKSIG;
block.hashPrevBlock = 0;
block.hashMerkleRoot = block.BuildMerkleTree();
block.nVersion = 1;
block.nTime = 1231006505;
block.nBits = 0x1d00ffff;
block.nNonce = 2083236893;
This scenario is unlikely because of the timestamp of the Genesis block (
block.nTime), which translates to January 3, 2009 at 18:15:05 UTC. Hearing about the Chancellor article a few days after it was published and then returning and setting the Genesis timestamp to a random hour from January 3, 2009 is highly unlikely.
Lastly, what if the article title on The Times’s website was later changed to include the Chancellor’s name, but initially didn’t and was the same as the one in the Genesis block? This is not possible since we found posts (here and here) from January 3, 2009 that reference The Times article from their website with the title as it can still be found there.
It is impossible to say, with absolute certainty, where Satoshi lived when he was working on Bitcoin – at least not with the available data. However, we can say, with reasonable confidence, that he was located in London. Put together, his writing style, his activity pattern, and the Genesis block message point to the capital of England as the most likely candidate.