Sections

Services

Information

US United States

Being angry on Twitter causes heart disease: Big Data fact or fiction?

Angry Tweeter.

A spate of articles have reported a study that claims the ability to predict a communities’ risk of heart disease from the language most commonly used by that community on twitter.

Analysing 148 million tweets, the researchers from the University of Pennsylvania and the University of Melbourne mapped the tweeters to counties in the US for which they also had data about the incidence of heart disease and other health and demographic data. The researchers found that they got a higher correlation between the language used in tweets from people in those counties and heart disease than the correlation between heart disease and other indicators such as income and education, smoking, chronic disease and racial background.

The language that signalled the increased likelihood of heart disease was broken down into 3 different categories of hostility/aggression, hate/interpersonal tension and boredom/fatigue. Words that were frequently used within those categories included words such as “fuck”, “hate”, “drama”, “bitches” in the hostility/hate categories through to “tired” and “shower” in the boredom category.

People tweeting in areas that had a lower risk of heart disease on the other hand tended to use words such as “conference”, “student”, “holiday” and “faith”.

The mainstream press, encouraged by the researchers involved in the study, have picked up on this study as another example of how “big data” and social media can reveal a whole range of information about a population’s health and well-being. One report featured the graphic from the study that showed the actual incidence of heart disease in a part of the US with that predicted by Twitter.

Tweets and heart disease

The interesting thing about the graphic is that if you actually compare the specific areas in each of the two diagrams, they are not particularly the same at all.

The graphic is a great example of how data visualisation can be used to mislead the viewer. The two diagrams seem similar because they use the same colours and have identical areas which have to be the same because no data is available.

Pick any two identical blocks however and the chances are that they won’t be the same colour. This is because the predictive power of the twitter analysis is actually quite low. Granted, the analysis may be slightly better than using a combination of other indicators such as income, education, chronic disease and race, but that difference is in actual fact minuscule.

So what is the general language on Twitter actually saying? Well, it is likely that the language used in tweets may reflect the socio-economic level of a particular area. But even this claim is making a number of unsubstantiated assumptions. It is easy to suppose that lower socioeconomic populations would tweet more frequently about things they hate using swear words than people who are more likely to be educated from higher socioeconomic backgrounds. Interestingly however, the study claims to have “controlled” for socio-economic status, which means that the researchers believed that the language used was not simply a representation of this status. Unfortunately, no details are given of how they did this and so it is not possible to say whether this was actually a reasonable assumption.

The study is interesting in that it suggests that a specific communities’ makeup can be identified by the language that they use on social media. This in turn may be an indicator of other factors such as the specific community’s health and well-being and the consequences of that health and well-being.

However, as attractive as those suggestions may be, the study showed weak correlations and gave very little underlying support for any particular theory of “causation”. So as much as they could speculate, the researchers could really not say why they saw these statistical results. The paper actually highlights some of the fundamental problems with so-called big data where large numbers distort statistics to point at imagined relationships.

What this study definitely does not say is that being angry on twitter will lead to, or is anyway related to, heart disease, which has been the unfortunate suggestion in the selling of the study in news reports.

Zombie technology: the technologies and companies that refuse to die

Zombie Worker.

When technology, and the companies behind it, fails, the end can come in a number of different ways. A technology can be mercifully put down, as with Google’s failed hardware media player, the Nexus Q. Alternatively, a failing company can be bought and shut down, as in the case of the once famous personal digital assistant maker Palm, who were bought, and then shut down by HP. Failing companies can also enter a more indeterminate, zombie state where the company may still earn enough money to stay open, but the company itself, and the products they produce, will never again be a significant force in the technology landscape.

Recognising a zombie company

Recognising zombie companies and technology is relatively easy. Companies with a languishing share price that shareholders are clearly only holding onto because they hope the company will be bought, is one clear indicator. Blackberry’s shares for example, popped 30% on the rumour that Samsung was about to buy the beleagured mobile phone company. The shares crashed back to their original value after the company denied the reports. Twitter and Yahoo also both benefited from the suggestion by ex CEO Ross Levinsohn that they should merge.

The fact that the market should respond to these types of rumours are clear signs that the companies have exhausted the option of developing their own products to continue making them relevant or competing against the market leaders.

Discussions about the death of a company or technology

Another indicator of a zombie company are the number of discussions that occur about whether the company/technology is actually dead or whether it will see a resurrection. This is being played out right now after Google’s announcement that its much maligned smart glasses were being pulled from public sale. Commentators are divided as to whether this signifies the complete death of the product or merely a pause before some form of re-launch. Google Glass has become a zombie product because even if it does survive, it will never have anything other than marginal interest.

In another case, reviews of BlackBerry’s latest phone, the Passport have tried to imply that this will somehow reverse its fortunes. Others propose growth for the company through services rather than hardware.

The key thing for zombie companies however is not to confuse the ability to stay in business with the fact that the business is actually viable. In the UK in 2013 for example, there were approximately 160,000 companies that were capable of staying afloat because they could pay the interest on their loans but had no way of ever being able to pay back the actual loans themselves.

Companies like Twitter for example, who are as yet to make a profit from anything other than the selling of their shares, can keep going on their IPO proceeds and by convincing people to invest further on the basis that they will eventually make money. The interesting thing with Twitter is that there is the belief that it can still make money somehow, with the right management. There are increasing calls for the CEO Dick Costolo to resign even though it may simply be that there is no viable way for Twitter to make enough money from its social network.

Zombie technologies

Zombie technologies pose a greater problem than zombie companies because it covers everyone involved in that technology. Zombie technologies are interesting because they often result from over-hyped expectations about their significance leading to a gold-rush surge of companies trying to catch the early wave of expectation. Massive Open Online Courses (MOOCs) for example, were going to transform the higher education sector by offering high-quality, free, online courses to the world. Companies like Coursera are still going only because of the large amounts of money that they have raised from venture capitalists. Unfortunately, the higher education industry proved resilient to change and Coursera’s attempts to make money out of ongoing professional education is never going to realise the ambitions of their investors. The same outcome is true for other MOOC companies like Udacity and edX.

Another topical zombie technology are crypto-currencies like Bitcoin. Bitcoin’s 80% fall in value since its peak in the past year has cemented its general failure to gain acceptance by governments, the financial sector and the public at large. This doesn’t mean the end of Bitcoin as there will be fringe uses for this technology supported by a core group of loyal fans. Its zombie state however will continue to be confused with a technology simply waiting for the right market opportunity to become the basis for the world’s future digital economy.

Zombie companies present a real problem in that they lock in funds, and employees who could otherwise be working more productively within their own startups or other companies.

Of course, eventually companies will stop trading, or be bought for their remaining assets, but that time may be surprisingly far into the future.

Bitcoin’s continuing price fall unmasks its underlying flaws. Is this its end?

Bitcoin’s future.

Bitcoin was dubbed the worst investment of 2014. As predicted however, 2015 has seen the continued fall in value of the currency that was supposed to fuel the digital age. In the last 10 days alone, it has lost 26% in value.

Bitcoin’s decline

If 2014 was a bad year for the digital currency, 2015 looks like it will be even worse. Barely days into the year, UK-based Bitcoin exchange Bitstamp was “hacked” and 19,000 Bitcoin stolen. At the time, this loss was valued at US $5 million. Bitstamp has since come back online, with revamped security from BitGo. It may however, all be a bit too late.

Hacks of Bitcoin exchanges have come to characterise the Bitcoin world. It isn’t something that is necessarily inherent in Bitcoin itself, but more of a feature of the types of companies that have sprung up around the troubled technology. At best, the hack of one-time leading Bitcoin exchange Mt Gox, was a result of sloppy coding and business practices. At worst, it was an inside job, defrauding its customers of $487 million.

A more ominous problem has cast its shadow on the future of Bitcoin. Bitcoin relies on people to engage in “mining” to validate every exchange of the virtual currency. Miners, do some agreed calculations, and if they are fast, or lucky, enough, will succeed in winning some newly produced Bitcoin in exchange for adding the transaction onto the Bitcoin ledger called the Blockchain.

The strategy of mining has become Bitcoin’s achilles’ heel. The design of Bitcoin dictates that the difficulty of mining will increase as more Bitcoins are produced and more miners get involved. This has led to mining being dominated by companies that can scale to the point where they can guarantee to earn a certain percentage of Bitcoins created each day. As Bitcoin’s value has dropped, the economics of the mining operation have changed, to the point that mining ceases to be economically viable.

Cloud mining company CEX.io suspended their mining operations this week, declaring that it needed the price of Bitcoin to be at least $320 before it would be able to resume its operations. Unfortunately for them, the price has dropped even further since and the likelihood of it climbing back to $320 seems slim.

Another mining company, CoinTerra, is being sued by a data centre provider for $5.4 million for unpaid fees. The cost of power alone to run CoinTerra’s services was $12,000 a day.

The underlying protocol of Bitcoin does allow for the relative difficulty of mining to be eased if it becomes to hard for miners to stay in operation. In fact, this happened last month for the first time since 2012. It could theoretically continue to become easier as the Bitcoin price drops. The issue is however, that this wasn’t supposed to happen. Bitcoin’s price was supposed to keep increasing as more Bitcoins came onto the market.

Bitcoins value relies purely on the belief of the people who buy and sell it. There is no central bank or government around to support it in the case of its value crashing to zero. Once that belief is questioned, Bitcoin becomes unsustainable. Even if the price of Bitcoin doesn’t go to zero, the chances the Bitcoin community convincing the wider public, governments, and industry that Bitcoin really represents the future of the world’s digital economy will become extremely unlikely.

For the time being, Bitcoin still has enough devotees who believe that the currency will eventually recover and still claim the crown as the future enabler of all digital commerce. However, even they are having their doubts that this grand technological experiment may have run its course.