Schrödinger’s Musk
Known for his vocal support of Elon Musk on X, Adrian Dittmann has become the subject of widespread debate. Here we examine what we can learn from account behavior.
Known for his vocal support of Elon Musk on X, the username AdrianDittmann has been the topic of debate on and off since 2023. Is he a fervent Musk fan, an AI clone, or could he be Musk himself? The controversy stems from Dittmann's voice and mannerisms in Twitter Spaces. He sounds uncannily similar to the CEO of Tesla and SpaceX.
For this edition of Trendy Context, an irregular blog that brings context to off-topic trends, let’s look at Schrödinger’s Musk.
What’s New
As of right now, there are a few findings that I’ve not seen reported elsewhere:
Analysis of the applications used to post by AdrianDittmann and ElonMusk.
Evaluation of interactions between AdrianDittmann and confirmed Musk burner accounts.
Timeline activity comparisons.
Burner Accounts
According to the Internet Archive, the AdrianDittmann account has existed since 2021—contrary to viral videos claiming it began in 2024. In April 2024, a lawsuit exposed two of Musk's burner accounts, as reported by Miles Klee in RollingStone on April 8, 2024:
Musk confirmed, too, that one exhibit entered into court records showed another account he operated for “test” purposes. The profile, Ermnmusk, came to light a year ago as a probable secret Musk account because he tweeted an image showing himself logged into it, and Motherboard then reported on a number of indications that it was likely his.
Two days after this information became public, AdrianDittmann interacted twice with one of the accounts. These are the only interactions between AdrianDittmann and the burner accounts in the collected data.
If interactions with a burner account predated the public disclosure, that might be significant, but they occurred two days after the information became public.
The ErmnMusk account was logged into Musk’s Twitter app on his iPhone screenshot.
Dittmann’s account has posted over 12,696 tweets from an Android device. The account has posted with an iPhone five times, and these were in 2023.
One might argue that exposure on another account led Musk to be more careful with other accounts, but even Musk cannot go backward in time (at least not that we know of). AdrianDittmann consistently posted from the Android device from 2021 to 2024, trying a few different posting interfaces but always returning to the Android.
Comparing Musk’s Wake-Wake Cycle to AdrianDittmann
There is also the matter of their sleep-wake cycles, or lack thereof, in the case of Musk.1 Musk’s sleep-wake cycle has grown less predictable over time. His graph does not drop down to nearly zero in the way that AdrianDittmann’s does.
Given the difference in tweet volume between the accounts, I examined a more extended timeframe for AdrianDittmann's tweets.
The accounts differ in patterns of activity. AdrianDittmann is least active at hour 16, one of Musk's more active periods.
While Musk's activity never wholly drops to zero—suggesting an irregular sleep-wake cycle—it reaches its lowest points during hours 9, 10, and 11, after peaking around hour 5.
AdrianDittmann shows a slight activity spike around hour 11 in the longer-term graph, though this difference may be less significant given the disparate periods covered.
Posting Interface
The data show AdrianDittmann posted 13,577 times between July 31, 2021 and Dec 30, 2024. "Twitter for Android" was the most popular interface used by the account with 12,696 instances, followed by "Twitter Web App" with 736 times, and "Twitter Media Studio" 124 times. The account minimally used "TweetDeck Web App" (12), "Twitter for iPhone" (5), and "BlackMagic.so" (4). The five tweets posted from an iPhone occurred in December 2023.
In comparison, Elon Musk's posting data shows a strong preference for "Twitter for iPhone" with 21,621 posts, while "Twitter Web App" accounts for only 41 posts and "Periscope" for four posts. Musk has never posted using Twitter for Android, Twitter Media Studio, TweetDeck Web App, or BlackMagic.
Under normal circumstances, the different posting patterns could be a point in favor of this being separate individuals. While people running multiple accounts that are not part of a large operation seem to stick to the same apps, this evidence alone is inconclusive. We must consider it alongside other factors. For example, even if two accounts use an iPhone, their distinct posting schedules could indicate different users.
Account Activity
The AdrianDittmann account has been more active in the past. Data show activity spikes coincided with other controversies where the account was allegedly “unmasked” as Musk. Even after Musk purchased the platform, AdrianDittmann didn’t consistently reach more than ten daily tweets until May 2023.
In contrast, Musk’s activity has steadily inclined over time. Time wrote about Musk’s posting habits, saying:
Mr. Musk posts vastly more than he used to. From December 2013 to the middle of 2018, he tweeted just over a dozen times a week, on average. Between then and October 27th 2022, when he completed the purchase of X, he was posting 50 times a week. Since the takeover, that has risen to around 220 a week.
Musk reached 252 posts on December 20, 2024, meaning the upward trend may continue. The trendline for both accounts from March 2024 to December 2024 shows an upward trend for Musk, whereas AdrianDittmann is a flatline.
Content
Using an NLTK Tokenizer, I split the tweet text into individual words (tokens).2 This social media-specific tokenizer is designed to handle social media posts, including emojis, hashtags, and mentions. The tokenizer processes each post's text and creates a list of lists, where each inner list contains either all tokens from a post or tokens grouped by sentence. Common English stop words (such as "the", "is", and "at") were removed to focus on the substance of the content.
Elon Musk's most-used terms include "😂" (2,502), "💯" (823), "🔥" (681), "cbdoge" (571), and "🤣" (560). Adrian Dittmann's most-used terms include "🤣" (1,343), "🧡" (852), "elonmusk" (819), "😂" (542), and "satributes" (392). While both accounts use laughing emojis frequently, they show different preferences - Musk uses the upright laughing emoji (😂) more often, while AdrianDittmann favors the tilted version (🤣). Neither makes frequent use of hashtags.
Identifying Speakers by Sound
Elon Musk's distinctive speech pattern includes hesitations, stutters, and occasional difficulty articulating thoughts, reflecting his unscripted speaking style. In contrast, Adrian Dittmann delivers speech more smoothly and fluently, albeit in Musk’s accent.
While modern technology could explain these vocal differences through AI-generated or masked speech, the speaker's natural variations in pace and emotional responses suggest human speech rather than AI.
Speaker identification relies on things like unique speech rhythms, intonation, and pauses that form recognizable vocal signatures. Skilled impersonators can mimic these predictable traits. The more distinctive the voice, the easier it becomes for listeners to recall and for impersonators to replicate its key features.
Impersonators who emphasize recognizable traits can create convincing imitations despite imperfections. The brain seems to simultaneously interpret who is speaking and what they are saying.
Another consideration, given what we know about mental shortcuts, is that when AdrianDittmann criticizes unions and praises Neuralink as he did in a 2023 Twitter space, it may increase the listener's subjective perception that he is Musk, even if he is not and even though the information about these stances is publicly available.
Familiarity with a person’s voice and accent may affect our accuracy when identifying a speaker. How well each speaker knows Musk’s voice would similarly affect their accuracy in assessing whether this is the same person, but when we read many people saying they believe it's the same person, we don’t know how many are familiar enough with either voice, even to be worth considering.
Well-Known But Indeterminent Facts
Several widely discussed facts contribute little to resolving whether Dittmann is Musk's burner account. For instance, both have appeared simultaneously in X Spaces discussions, but it's common for people to host a space from one account while speaking from another. It's not hard to do.
Similarly inconclusive are Dittmann's apparent "slip-ups," where he closely mirrors Musk's mannerisms and views, occasionally speaking about Musk in the first person. These apparent Freudian slips have been featured in viral posts that assert this proves the speaker is Musk himself.
The speculation drives traffic to Dittmann's spaces, boosting his engagement and potential earnings. Thus, one could argue there is a monetary incentive to keep the guessing game alive. To be clear, the account frequently vehemently denies being Musk and has posted about frustration with the inability to dissuade people of this.
AdrianDittmann has gained thousands of followers in the past week, meaning if it is a strategy to maintain audience interest (though I do not assert it is), it’s certainly effective. The question has come up more than once in most, if not all, of the spaces I listened to, and it’s easy to understand why: he does sound a lot like Elon Musk.
Verdict
If this were any other case, considering the account data, I would say this involves two individuals. The account data suggest that, in my opinion. Still, given the number of potential variables and Musk's extreme outlier status, I remain undecided and open to new information.
I would be more surprised if it turned out to be Musk, given the amount of time the other account spends in Twitter Spaces, but that's me making assumptions about how busy the CEO of multiple companies might be. Personal impressions can be easily manipulated and mislead us.
I'm not overly familiar with either of the voices nor am I a native South African English speaker. That, coupled with the potential effect of what he is saying, means my feeling that they sound like the same person is probably less significant than it feels.
Subscribers have access to the complete dataset. Please do not share it publicly; if used in reporting, please cite us.
Keep reading with a 7-day free trial
Subscribe to InfoEpi Lab to keep reading this post and get 7 days of free access to the full post archives.