As Bitcoin has grown from a niche subculture into a global financial phenomenon, mainstream media outlets have grappled with what the cryptocurrency is, and how they should cover it.
The results weren’t always pretty, especially at first. Crypto enthusiasts have often complained about media bias. For example, here’s a 2018 reddit thread in which the original poster argues the media is biased against Bitcoin, and every commenter in the thread agrees. Here’s a very long BitcoinTalk thread from the same year, in which most commenters also say they see anti-Bitcoin bias.
Certainly, it’s not difficult to find examples of negative Bitcoin coverage in the press. But is the media really biased against Bitcoin? Is its coverage out of step with the public’s interest? Are some outlets better than others?
To find out, LongHash took a deep dive into several years of mainstream reporting history.
Quick Notes on The Data Set
The data set we used contains the content of 2.6 million articles, news published by 26 different high-profile outlets between January of 2016 and April of 2017. Outlets include major news networks like CNN and CNBC, major newspapers like the New York Times and the Washington Post, as well as magazines and a variety of popular news websites like Wired, TechCrunch, Gizmodo, Vice, and Mashable. This data set does not include media outlets that primarily cover crypto.
How Much Does the Media Cover Bitcoin?
Of the more than 2.6 million articles archived in the data set, 3,580 of them mentioned Bitcoin in the headline. While that may seem miniscule, the data set contains articles about everything. Compared to “US dollar” (1,368 articles) and “Ethereum” (282 articles), for example, Bitcoin seems to have done quite well.
Unsurprisingly, our analysis found that Bitcoin coverage tracked pretty closely with Bitcoin’s price. Coverage spiked along with the price in 2017, and over the time period covered in the data set, the number of daily Bitcoin stories showed a moderate correlation (0.39) with the BTC price, according to a Pearson correlation analysis. (Pearson correlation scores range from 1, a perfect positive correlation, to -1, a perfect negative correlation).
If we chart this data, though, we can see that even though Bitcoin’s price fluctuated dramatically at times after the 2017 bull run, media coverage dropped down to near its pre-bull-run levels. Another price spike in mid-2019 didn’t spark the same level of media interest.
When we checked the media coverage trends against Google Trends search data for “Bitcoin” over the same period, we found they tracked remarkably closely — 0.88, a very strong positive correlation.
Knowing there’s correlation doesn’t necessarily tell us anything about causation. It’s possible that media hype is the primary driver of Google search trends, and it’s also possible that public demand for information about Bitcoin is driving media coverage. When we lined up the data using precisely the same dates, though, we found that search interest tends to spike before mainstream media coverage does.
In other words: It appears that mainstream media coverage is following public demand for Bitcoin stories, rather than the other way around.
Of course, not all outlets covered Bitcoin the same way. We analyzed all of the articles from each outlet to determine what percentage of its total coverage was dedicated to Bitcoin.
What we found wasn’t particularly surprising. CNBC, the most finance-focused outlet on the list, dedicated the highest proportion of its articles to Bitcoin. It was followed by tech-focused blogs Gizmodo, Wired, and TechCrunch. Mainstream outlets like CNN and the New York Times fell somewhere in the middle. Sites such as TMZ and Refinery 29 didn’t dedicate much time to Bitcoin, which makes sense — they don’t aim to provide comprehensive finance coverage.
How Objectively Does the Media Cover Bitcoin?
Of course, when and how much Bitcoin gets covered are only part of the story. The bigger question is how Bitcoin gets covered. Is there really an anti-Bitcoin bias?
To find out, we ran the 3,500+ Bitcoin articles through two different sentiment analysis tools: VADER and TextBlob. Although they work slightly differently, both analyze sentiment primarily by looking at word choice. Both output their results in the same way: a numerical score between -1 (completely negative) and 1 (completely positive) for each article.
For example, givingTextBlob the sentence “What an absolutely great day, I love it!” results in a score of 0.71. “What an absolutely horrible day, I hate it!” gets a score of -1. While this kind of machine assessment is far from perfect, it allows us to get an idea about the sentiment of written text without having to read and assess thousands of articles individually.
TextBlob also attempts to assess the subjectivity of an article, scoring this between 0 (completely objective) and 1 (completely subjective).
When we ran the Bitcoin article text through these two tools, they scored them quite differently, but neither found evidence of a negative bias against Bitcoin.
VADER’s analysis returned a broad range of scores — each dot represents an individual article’s score — but the largest clusters fell in the top half of the graph, closer to 1 (very positive) than -1 (very negative).
TextBlob’s sentiment analysis produced a much narrower range of scores, but the same general trend, with sentiment tending to lean more towards positive than negative.
TextBlob also found that most articles tended to fall somewhere in the middle of subjectivity and objectivity, although there are certainly a few very subjective outliers (the smattering of pink dots near the top of the chart):
We also looked at sentiment scores by outlet to see if specific outlets tended to be more or less bullish. For this analysis, we looked at average sentiment scores across all an outlet’s Bitcoin articles, so we included only outlets that had published 20 or more Bitcoin-focused stories.
While there was plenty of disagreement on the specifics between VADER and TextBlob, we can see again that in both assessments, all outlets had a positive average score.
According to both tools, TechCrunch, Vox, and the New York Times are among the most bullish outlets. Reuters and Axios scored towards the bottom of the pile in both analyses, but with none of them dipping below zero, it would be unfair to label any of them as particularly bearish — at least according to this analysis.
TextBlob’s subjectivity analysis is also interesting, although again, the scores are fairly close. As one might expect, traditional journalistic outlets that value objectivity scored lower — they were more objective. Internet-based new media outlets such as Mashable, Vice, Vox, and Gizmodo landed on the more subjective side of the spectrum.
So, is the media biased against Bitcoin?
We didn’t find any evidence of major media bias against Bitcoin in our analysis. The data show that media coverage follows public demand for Bitcoin information pretty closely, and two different sentiment analysis tools failed to turn up any evidence of anti-Bitcoin bias at any of the outlets that cover it regularly.
Of course, these conclusions come with quite a few caveats. The data, while expansive, is still limited. It covers just 26 outlets over only a few years, but there are many more media outlets writing about Bitcoin. Machine-based sentiment analysis is far from perfect, and there are other natural language processing tools that, if applied to this same data set, might produce different results than TextBlob and VADER.
It's also important to remember that this kind of analysis is based primarily on word choice. It assesses the sentiment and subjectivity of language, but it cannot access other important factors like factual accuracy.
The best way to assess any crypto news article is still with a healthy dose of human critical thinking. But the next time you see a post about the mainstream media being a bunch of Bitcoin bears, you can probably shrug it off. While there are definitely some negative people out there, on the whole, it seems like the media is actually fairly bullish about Bitcoin.
