Blog archive


Max St John

What happens when we meet the bots?

Just over a week ago, trading firm Knight Capital lost $440 million in the space of a few minutes. This was all thanks to an upgrade to their trading software that went very wrong, resulting in bad trades that cost the company around half its equity.

It’s not the first time this has happened – in 2010, nearly $1,000 billion in value was wiped off the Dow Jones industrial average in a few minutes, thanks again to high frequency trading ‘algo bots’.

This sent the markets into panic and triggered a frenzied clean up that narrowly avoided a contagion effect that could potentially have created a global economic meltdown.

These algorithmic trading bots are thought to make up around 60% of high frequency trades on major stock exchanges, and up to 40% of trades are thought to be between bots – all “making a fraction of a cent at a time, multiplied by hundreds of shares, tens of thousands of times a day”.

Aside from the scary thought that this is still happening despite warnings from experts (“We will be vulnerable until we learn that deep, liquid and fair markets will never result from unrestrained and unregulated high-speed algo-bots.” said a comment piece in the FT over a year ago), I thought the idea of bots, silently trading between each other in the ether, reacting as new algorithms enter and exit the market place, was pretty mind-blowing.

Then I found out about quote spam – where the bots make thousands of seemingly pointless quotes designed to add noise to the market, cost competitors through having to store and analyse the data, trick them into revealing information or triggering trades. This has dramatically increased over the past five years, as shown by real-time data service Nanex in the gif below.

Nanex visualisation of rise in quote spam

I’ m not sure if Ross’s obsession with robots has rubbed off on me a bit but the idea of the algo-bots and their dark and silent meddling in our world from trading cyber space got me thinking about what happens when the bots and us cross paths more overtly.

Of course we’re surrounded by bots that crawl, click and catalogue content and advertising, and have been for a long time, but I realised that the bots are already very much among us.

A recent study showed that nearly half of some big brands followers are bots and if you’re on Twitter you’ve probably had some experience of certain types – but aside from the random, link firing or those listening for specific keywords to tweet back a link to a ‘relevant’ product or service, there’s another type of twitter bot I found interesting.

These sit out there tweeting a seemingly human stream of quotes and pithy comments that presumably go unnoticed by both Twitter and its users – simply to create a home for a link to bolster someone’s SEO rankings.

It didn’t take me long to find one: ‘Laura Bolt’ (it didn’t tweet links when I looked, but it does link to something racy looking in the bio, – if you click something don’t blame me).

She’s supposedly from New Mexico and has been on Twitter since 12th March 2012, pushing out a mix of pensive observations on life and slightly naughty pick-up lines.

But she’s far from alone – do a quick search for the text in one of her tweets and you get a sense of the scale of what’s happening now - hundreds, possibly tens of thousands of Lauras, with their slightly-too-polished profile pictures.

Twitter search results showing spam bots

The scary thing is: Laura gets retweeted by some of the real people in amongst her 423 followers (presumably some bots, some people who simply follow back). Seemingly undetectable to Twitter’s sophisticated technology and even the brain of human beings, the bots have infiltrated our world.

Laura and her bot friends are arguably harmless and clearly pretty simplistic, but where could this end up?

What if networks of bots, instead of churning out pointless quotes, made allegations about politicians, expressed seemingly real product recommendations, subtly influencing the we vote, the brands we favour, or the charities we trust?

If they can manage to go undetected now, soon they’ll be able to engage in complex interactions between themselves, to make their behaviour look even more trustworthy and believable, or even with you and I.

All it takes is for the bots to understand what makes us tick and who we listen to – relatively simple things to do using conversation and network analysis – and our worlds could be inextricably linked.

This post was filed under Innovation, Social media, Technology, The future and tagged , , , Comments are currently closed.


  1. felix

    Since bots manipulate the stock market and some government policy is dictated by the market, you could argue that bots already control the world to some degree.

    Stock market volatility will increase as more trades are made by bots. I expect a lot more ‘flash crashes’ and recoveries in future. Is this a bad thing? I expect we will just learn to deal with it.

    Posted 14th August 2012 at 7:59 pm | Permalink
  2. Both really good points Felix – I guess whether it’s a good or bad thing that these bots are having more impact depends on how it affects you. From my poking around, it looks like there are a significant number of people calling for tighter regulation and control of bots and HFT, but there are probably an equal number on the other side of the fence too.

    Thanks for the comment.


    Posted 14th August 2012 at 8:40 pm | Permalink
  3. Interesting stuff. I’m already disturbed at how much more specific those bot comments that appear on blogs and elsewhere are, than they were just 18 months ago. And I’ve certainly been fooled into retweeting fake news accounts (fortunately not the lovely Laura though).

    Posted 15th August 2012 at 11:41 am | Permalink
  4. Interesting…

    On the one hand it’s disappointing to think that some of those followers I have, usually with a cleavage heavy picture, might not actually be there for the depth and insight of my twitterings,

    But – there is clearly a chronic lack of public debate about the implications of this explosion / permeation of AI (or whatever one wants to call it) into countless areas of our lives. Set against emergence of things like pilotless drones in combat zones, and run away developments in nano-technology, it really does begin to feel like the back story to The Matrix, or T2…

    In fact, fiction seems to be one area where these ideas / questions are thrown around a bit. Charles Stross is interesting and the recent Robert Harris novel is hung on a algorithm driven software system going feral on the world financial markets and almost bringing it all down… An odd premise for a thriller (and pretty trashy I’m afraid to say) but entertaining, and already optioned by Hollywood.

    Posted 15th August 2012 at 11:44 am | Permalink
  5. Tim – thanks for the comment – I completely agree, it’s becoming increasingly difficult to tell bot from not-bot, especially when our attention is already spread so thinly.

    Tom – Robert Harris’s novel sounds very close to the potential near-future if some people I came across are to be believed. Would be interesting to have a public debate about the potential negative impacts of bots expansion (further integration?) into our world but equally, I wonder how these social bots might have a positive role to play?

    Posted 15th August 2012 at 3:41 pm | Permalink
  6. This just reminded me of this blog post I wrote back in 2009. I set up a test twitter account, tweeted ‘do not follow me back’, followed 1000 people and watched what happened. Three years later and it has over 1000 followers. Most of them LOOK real…

    Posted 15th August 2012 at 5:15 pm | Permalink
  7. Beth – really interesting, and sounds similar to an experiment Dirk Singer (the Rabbit Agency), recently ran:

    Posted 15th August 2012 at 5:36 pm | Permalink
  8. Thought people on this thread might be interested to see how this has actually been done. A Google+ product manager created a Twitter bot that built trust with a group of people, to the point where they were concerned about “her” well-being following a “running injury”. Also – activists in Mexico think that political parties are already trying to influence voters via Twitter bots:

    Posted 21st August 2012 at 9:14 am | Permalink
  9. Great post and highlights some really interesting implications for brands/media in terms of signal/noise ratio. I’ve always said I’d *love* to see a big brand’s Facebook Page fan numbers once spammy/agency/industry fans had been removed!

    Additionally, did you see Rory Cellen-Jones’ blog post about Facebook spam recently?

    And finally…… not sure if you;ve come across it, but the London Review of Books had a good essay on algorithmic trading and it’s results last year:

    Posted 25th August 2012 at 9:18 am | Permalink
  10. Cheers Simon – Rory’s post made me wonder if this kind of ‘problem’ would move the big platforms towards supporting a single citizen digital ID (for better or for worse). The LRB essay looks really interesting – I’ll check it out.

    Posted 25th August 2012 at 3:06 pm | Permalink