How dead is the internet?

This BBC podcast is supported by ads outside the UK.

At Azure Standard, we believe healthy food shouldn't be complicated.

That's why we've spent 30 years delivering clean, organic groceries and earth-friendly products to families who care about what goes on their plates and into their lives.

From pantry staples to wellness essentials and garden-ready seeds, everything we offer is rooted in real nutrition, transparency, and trust.

Join a community that believes in better naturally.

Visit Azure Standard.com today and discover a simpler, healthier way to shop for the things that matter most.

A happy place comes in many colors.

Whatever your color, bring happiness home with CertaPro Painters.

Get started today at Certapro.com.

Each Certipro Painters business is independently owned and operated.

Contractor license and registration information is available at Certipro.com.

Hello, and thanks for downloading the More or Less podcast.

We're your weekly guide to the numbers in the news news and on the internet.

And I'm Tom Coles.

In the early 2020s, a conspiracy theory started circulating online.

It's called the dead internet theory.

Sorry, not sure what came over there.

Anyway, this theory suggested that instead of a vibrant digital super community where people freely share things like cat videos and conspiracy theories, the internet was instead basically dead.

An AI dystopia controlled by the deep state, where almost everything you see and interact with is generated by computers.

Now obviously the internet isn't totally dead.

You can prove that wrong by posting something nice about more or less on Facebook if you like and asking a friend to take a look.

But the theory does hint at something real.

The internet is full of things that are not alive.

Things called bots.

And today we're going to look at one particular bot-based claim, as seen in this headline.

More than half of all internet traffic is now bots.

The internet might not be fully dead yet, but this stat makes it sound like it's certainly dying.

So,

is it right?

First things first, what is a bot?

Gareth Tyson is a professor of computer science at both Queen Mary University of London and Hong Kong University of Science and Technology.

So a bot is an umbrella term that typically refers to any type of autonomous software that's connected to the internet and then can interact with other systems or other users.

And it's from the word robot, presumably.

Yes.

Okay, so you might be thinking that this statistic, that 50% of internet traffic is now bots, is telling you something important about social media posts and chat room conspiracy theories, about content that you might actually see when you're browsing the internet, and how much of that is being generated by non-human bots.

Is that what this stat is talking about?

I don't think so.

So that type of activity is one example of a bot activity, but it's not the only one.

The 50% figure comes from a company called Imperva, who sells software that's designed to stop some of the many millions of bots that are cruising around the internet.

We'll get to the social media bots you might actually see in a Mo, but Imperva do not have any way of counting them, and so they are not what this number is referring to.

Instead, the bots in question are doing other things.

Probably the most common type of bot would be one that participates in something called a botnet, which is a network of bots that work together.

And often these are up to relatively malicious activities.

Denial of service is one example.

So imagine you're a competitor and I want to take your website down.

I would employ the services of all these bots to visit your web page repeatedly and overload it so that customers, when they try to visit your page, can't access it and perhaps therefore come to my page instead.

One other type of quite common bot would be those that are trying to break in to people's computers.

They break into computers and get them to do, well sometimes, more denial of service attacks.

or they might steal your personal data or something else.

There are also bots doing various horrible sounding things like stuffing and scraping.

They're basically like the robot minions of nasty criminal or corporate masterminds.

But before the bot union complains, not all bots are bad.

Another really common example of a bot would be a crawler.

So these are bots that visit web pages with the interest of gathering data.

Services like Google do this every day.

So they visit web pages to collect data that then allow you to use their search engine.

So in terms terms of how general internet users might actually experience this, this isn't content that they're seeing.

In most cases, you would see the secondary consequences.

So for instance, if a botnet is attacking a particular web page, users might notice that because the web page becomes slower.

The 50% stat from Imperva is counting something most people never see directly.

It's also counting website requests, not volume of data, which makes the number sound really big.

What's more, I check with the company and they don't have any method for extrapolating the proportion of bots they know about from their software to the whole internet.

So it's really just a number for the amount of hidden bots that one company sees.

Bottom line is that there happens to be a botnet war going on in the background of the internet that most of us are gloriously unaware of.

And one company on the front line of that war, possibly unsurprisingly, came up with a big number for the number of bots.

This is a figure that tells you something about one company's perspective on the internet.

Is there a reliable figure for the amount of bots in the whole internet?

Not really.

And that's the case for several reasons.

The first and most obvious one is that we don't actually have a single comprehensive data set that covers all global traffic.

The internet is made up of thousands of completely independent networks, all of which are operating autonomously.

And there's no central point which connects them all together that allows us to actually measure all the different traffic flows.

The second challenge that makes it very difficult is the fact that we don't really have effective ways to identify every single different type of bot that's out there.

So, even if you could get access to the world's traffic data, it would be difficult to extract from that data every single case of a bot interaction.

The 50% figure isn't talking about the whole internet, and it does not include content on the bits of the internet we see and interact with through social media.

But that does not mean that social media is bot-free.

There are many examples of the types of things these bots get up to.

Probably amongst the most high-profile ones are things like election interference, where you employ an army of social media bots to try to promote your cause.

And if, for example, you're running an election, you might employ these bots to constantly talk about how wonderful you are and maybe share news articles that are very positive.

Another common use for bots is things like brand promotion.

There are bots out there using algorithms and AI to write posts and leave comments all over social media and also to like and share posts, videos, songs and anything else.

Wherever there's an incentive, financial or otherwise, there's a bot.

So for instance imagine you want to have a thousand followers on your Twitter account or you want to have a million people view your video.

You can go onto some services which employ bots on your behalf, you pay them some money, and then within a few days those followers will arrive or all of those views will arrive onto YouTube.

So in terms of the dead or aliveness of the internet, you'd want to know the number of bots on each different social media site.

However, that information is also incredibly commercially sensitive.

So there have been many different estimates, but one that might be relevant would be the estimates that came about during 2022 when Elon Musk was planning on acquiring Twitter, because this has real commercial interest.

Because if your social media website is 99% bots, then clearly it's less valuable to advertisers.

So at the point of acquisition, Elon Musk's team were estimating that 33% of visible accounts on Twitter were either false or spam bots.

Following that, there were various companies that were enrolled to try to estimate the figure independently, and a few different numbers came out.

So one of the most prominent one was an estimate of around 11%

bots.

Another one estimated about 5.3% bots.

So you can see that there's quite a lot of diversity in these different estimates.

And no right answer.

And no 100% provably right answer.

These companies X, Meta, Google, are in a continuous arms race with the bots, one that's getting more complicated by the second with the advent of AI.

There is a lot of bot in what you see online, but precisely how much?

Shrugging shoulders emoji is probably the answer to that one.

Assuming you're not a bot, please get in touch if you've seen a number in the news you think is worth looking at.

The email is more or less at bbc.co.uk.

Until next time, goodbye.