A Moment on Twitter

I recently found myself wondering what an average moment on Twitter would look like if you could take a snapshot of all the tweets being posted at any given point in time. Where would the tweets come from? What would they be about? Would they be mostly just text, or would many include photos and URLs to other sites? Would they be tweets sent out into the ether, or engagements with others such as replies, retweets, and quoted tweets? I don’t have access to the Twitter “firehose” but I found a way to take a decent snapshot of a moment in time for Twitter, and today I’m going to share what I found.


First, a brief word on methodology. Twitter provides access to its tweets – that is, every tweet passing through the service at any given point in time. As a developer, you can tap into this firehose to power an app of one kind or another. But as a regular user of Twitter, you really have no way to see the full stream of tweets at any given point in time, which means you have to try to find a proxy. In the end, the closest I was able to come was running a search of all tweets for a simple word: “a”. That doesn’t provide a perfect sample of tweets – for one thing, not every language uses that letter as a standalone word, though two of the world’s most widely spoken (English and Spanish) do, along with quite a few others. I scrolled down until the search results populated with a thousand tweets (posted over the course of around three minutes). That was the basis for my analysis. It’s certainly not perfect – for the reasons just mentioned, it’s skewed in terms of language, but I think the results are probably fairly representative nonetheless. Someone with access to the firehose and the right developer tools could obviously do a more thorough and statistically significant job on the analysis.


In terms of content, there was a massive range. There were tweets about politics (17 of the thousand mentioned Trump by name and six were in fact replies to his official account); tweets about sports, from soccer to the NFL; tweets about cars, celebrities, and even one about a Korean proverb. There was a fair bit of what might be called offensive language – at least a dozen terms some would consider offensive, though mostly not used in that way (which highlights one of Twitter’s biggest challenges with policing abuse); ten s-words and six f-words, and quite a bit more lewd, vulgar, and generally PG-13 or R-rated stuff. Three tweets mentioned sex using that word specifically.

Interestingly, though, the content was dominated by text – only around a tenth included photos, while around 160 included URLs. Among the photo tweets, a number were selfies, while a few more were screenshots of messaging exchanges, bits of text, or pictures of products from brand accounts or their fans. Of the text tweets, very few even approached 140 characters – I haven’t done an exhaustive analysis but I’d say the average length is probably under 100 characters, with many closer to 50. Among the URLs were many links to YouTube videos which users had either liked or, in a smaller number, posted themselves, 16 fb.me links to Facebook (and seven direct references to Facebook in the text of tweets). There were 11 Instagram links.

Among the thousand tweets, there were just two Twitter polls, both about politics (one about voting behavior in a recent election in a Spanish-speaking country and the other about raising the legal age for adulthood in France). There were 327 hashtags in the sample, including a handful in people’s names, with many others appearing in tweets where it seemed almost every word was a hashtag. The vast majority had none. A televised debate in France using the hashtag #LeGrandDebat generated 27 of those hashtags and this was a trending topic at the time (as were several other terms related to the debate or the French election).


One of the things I was most curious about was the level of interactivity on the service. Some 266 of the 1000 tweets were officially in reply to other tweets using Twitter’s reply feature, while a number of others were likely also manually-created replies or @-mentions. There were 52 retweets and another 50 or so quoted tweets. All of this suggests that, although a majority of the tweets were standalone and neither replies to nor regurgitations of others’ tweets, a good-sized minority were interacting in some way with other users or their tweets.


Interestingly, just five of the thousand tweets came from verified users, with the rest coming from ordinary users without the blue checkmark. I haven’t done a thorough analysis of all the users who posted tweets in this very brief window but looking at a sampling suggests that, on average, they follow about 700 accounts and have roughly the same number of followers. However, that average is skewed quite a bit by a small number of accounts with high follower counts. The median follower number is 237, which feels more representative of the accounts than the 700 number. A number of accounts had follower numbers in the teens and one had literally zero followers (indeed, the tweet from this account read: “I figured that since this is anonymous and dont have any followers, that I should maybe just use this as a way to talk about whatever I want”.

As I mentioned at the beginning, searching for the word “a” eliminates certain languages and countries where they’re spoken, but I still had tweets in English, Spanish, Portuguese, French, Italian, and even somewhat bizarrely, Arabic among the sample. A more representative sampling would no doubt have featured tweets in many other languages and from many other countries as well.


I found my experience looking through these tweets fascinating. In some ways, Twitter is the anti-Facebook: whereas the latter is by default private and it’s therefore quite difficult to learn anything truly personal about strangers’ lives using the service, Twitter feels downright voyeuristic by comparison – minutiae of people’s lives are laid bare for all to see (and to search using keywords and hashtags). The sheer variety of what people post is staggering. For someone like me, with a carefully curated list of accounts to follow, the experience of Twitter is very narrow by comparison – I follow mostly people who write about and work in the tech industry but that’s a tiny fraction of what’s out there on Twitter. I’m struck by the fact many of those whose accounts made it into the sample are following hundreds of others on the service, likely a mix of strangers and brands alongside personal acquaintances and, in many cases, are interacting with other accounts, sometimes politely and sometimes more aggressively. Twitter continues to be a unique window onto our world, with all its amazing facets, warts and all.

Published by

Jan Dawson

Jan Dawson is Founder and Chief Analyst at Jackdaw Research, a technology research and consulting firm focused on consumer technology. During his sixteen years as a technology analyst, Jan has covered everything from DSL to LTE, and from policy and regulation to smartphones and tablets. As such, he brings a unique perspective to the consumer technology space, pulling together insights on communications and content services, device hardware and software, and online services to provide big-picture market analysis and strategic advice to his clients. Jan has worked with many of the world’s largest operators, device and infrastructure vendors, online service providers and others to shape their strategies and help them understand the market. Prior to founding Jackdaw, Jan worked at Ovum for a number of years, most recently as Chief Telecoms Analyst, responsible for Ovum’s telecoms research agenda globally.

2 thoughts on “A Moment on Twitter”

Leave a Reply

Your email address will not be published. Required fields are marked *