Experimenting High Frequency Trading Principles on Twitter
A common issues is “how to attract people attention to my web page/twitter when I don’t have any reputation?”. Since we live in an attention economy (attention is a scare commodity), why not tackling such question under the prism of finance and applying High-frequency trading technics to gain attention shares. Here is a fun experiment.
(Article updated – original article published in 05-2007 for MyBLogLog service. Time flies – We call now such process Growth Hacking….
Here is my twitter account around the end of September: 151 followers, 143 followings and 64 tweets
And at the end of December: 1139 followers (+ 650%), 128 followings (-11%) and 87 tweets (+35%) after executing an algorithm using High Frequency Trading principles.
And it’s not fake followers as we can see sometime. (check it using the service http://fakers.statuspeople.com)
So how does it work?
High Frequency Trading in Finance
According to Wikipedia, High-frequency trading (HFT) is the use of sophisticated technological tools and computer algorithms to trade on a rapid basis.
An investment position in HFT may be held for only seconds, or fractions of a second, with the computer trading in and out of positions thousands or tens of thousands of times a day.
HFT aims to capture just a fraction of a penny per share or currency unit on every trade.Fractions of a penny accumulate fast to produce significantly positive results at the end of every day.
To really understand this concept and the context of the rest of the article , watch the Great TED Talk “how algorithms shape our world”
HFT in Attention Economy
Let’s now back to the question how can I attract people attention. To tackle such problem a first step is to understand that we live in an attention economy.
What is Attention Economy?
Attention economy treats human attention as a scarce commodity and applies economic theory to solve various information management problems. As pointed by Herbert A. Simon , in an information-rich world, a wealth of information creates a poverty of attention and a need to allocate that attention efficiently among the overabundance of information sources that might consume it”
We can extend it to the social sphere: the attention that one dedicates managing his/her interactions with others. I named it social attention in previous works. We thus have a theoretical cognitive limit about the number of people with whom one can maintain stable social relationships (Dunbar’s number – 150)
Once we understand that attention is the currency in this economy, why not applying financial technics and especially HFT principles to allocate and bargain such limited resource in the social sphere?
So I have a given social attention capacity that I want to invest in a social network using HFT strategies. To be aligned with HFT, I cut such capacity into a set of micro attention investments on which I expect to gain extra shares of attention.
In the twitter world, with a normal attention capacity, I can follow 200 twitter users (without taking into account the information bandwidth of each twitter account), that means I can invest my whole attention on 200 different micro-attention positions at a given time that may be held only for short period of time, my algorithm selling my attention (following someome) and rebuying it (unfollowing) several time per day. Of course by doing that I artificially increase my attention capacity .
Gain of Attention & Social Reciprocity
The goal is not to follow lots of people (it is the reverse cf ‘radar and invisiblity section’). It’s an investment from which I expect to gain new attention shares. An attention trading relies on psychology notion of Social Reciprocity. I’m sure you know the simple social mechanism used in websites like linkedin: if a user A visits a user B’s profile/blog, this one will know it. Because a user is always curious to know who is interested by him, he will certainly go to the visitor’s page. So B will certainly go to visit the user A’s page. This very simple feature has a real social impact on the people behavior and can be hijacked for our purpose by targeting people who are more willing to have a reciprocity behavior e.g. targeting the long tail of people having not so much people following them ( the large majority of the twitter users). Of course other human factors should also be taken (topics, social closeness).
Speed and Social Volatility
About the duration of the position (i.e. following someone) , in the case of social transaction, the speed can’t be at the millisecond as in finance. The trading won’t work. too fast for a social interaction. However paying attention for someone only 1day is socially quite fast and can be considered as a very volatile behavior at the social level, and thus is aligned with HFT principles.
Here is a sample of the impact of the algorithm. Around the 24th of november I (re)started the algorithm increasing my followers of +40 followers each day. I don’ have the curve of my followings but it will be quite constant, around 150 followings. I then stopped the algo (flat curve) at the beginning of december (maintenance mode).
FEb 2013: I also tried a version for Linkedin. worked well! See the pick of the number of people checking my profil when I started the algo – no aggressive mode (around 100 people / week) and just stopped it afterward ( well in linkedin the pb is actually a bit different / simpler so the analogy with HFT is not totally relevant, anyway)
Radar and Invisibility
One of the properties of HFT is its invisibility (cf watch the above Ted Talk). Exactly like HFT, this behavior is nearly invisible for a common user going to my twitter account. A naive/basic spamming behavior follows people and thus accumulate a large number of followings. The large # of followings makes their attention allocation suspicious. However If one go to my twitter account he/she will see a common number of followings e.g. 200/300. But these followings change every x hours. So detecting such artificial augmentation of my attention allocation is difficult since one has to identify high frequent tiny attention shifts done every x hours.
Volatile Attention: the devil ?
Such practices is a “destructive force” against the philosophy and real value of twitter which is real conversion since it does not encourage long-term attention investment to enable a conversion. My goal is thus not to encourage this practice. It was just a fun experimentation . I stopped the algo since december. (Maybe people who are reading this post have been attracted by my bot). However 2 remarks beyond naive rejections:
- Continuous partial attention. This strategy of volatility of attention is already present in our environment.As I previously mentioned, due to a growing connectivity, people are changing their strategies to manage their social networks. L.Stone called this emerging behavior “Continual Partiual attention” (CPA). It involves an artificial sense of constant crisis: people attempt to stay partially but continuously aware about the activity within their networks. maintaing a relationship with 1000 Facebook friends creates such volatility of attention. so while my strategy is more explicit, people are doing implicitly the same thing. this is the natural strategy in a rich-interaction world. My algo is just going a step further.
- As strange as it is, It triggered real appointments in the physical world. Some people contacted me because I followed them and were curious to know me better. So I had lunch with them. But this is another topic that I am currently investigating (random lunch). I will publish some findings about it on another post
PS: such kind of practice might not be new on twitter (I created similar stuff in 2007 for my mybloglog service), however reframing the problem by connecting the financial world and making an analogy with High frequency Trading is, as far as I know.