18 Things, Part 2

18 Things is a slightly different take on the usual “Annual Predictions” posts. Over the next few weeks I’ll be posting 18 things I believe might happen in 2018, with a brief overview of why. Each will also have a confidence level (an idea nicked from Scott Alexander), so I can come back at the end of the year and see how I went. I’ve mostly stuck to my circle of competence, but please don’t read in to anything too much, and definitely don’t go betting your house on any of these…


7: The beginning of the end of the cult of disruption

The boardroom buzzword of the decade is undoubtedly “disruption”. Businesses have been furiously working out how they can “disrupt themselves”, and watching in awe as the new disrupters disrupt entire industries with great disruption.

Any questioner of disruption is quickly labelled a Luddite, grasping hopelessly to the past. But increasingly we will realise that it’s important to ask the hard questions about where this is all headed.

Disruption was cool when it was happening to the hotels and the record labels and the taxi drivers. But now it’s happening to your job. And suddenly it’s not so cool.

An oft-cited study from Oxford suggest that 47% of jobs in the US are at risk from automation and computerisation. As this statistic becomes a reality for white-collar workers over the next year, it will become increasingly hard to find anyone who feels their career is secure.

(If you’re interested in more on this, I wrote this in 2015, questioning our disruption obsession.)

Confidence: 80% (but I’m relying on you to be honest on if you feel secure in your job)


8: A top 20 global advertiser pulls out of Facebook or YouTube

Over the past two years, brands have increasingly been questioning their ability to accurately measure and verify digital advertising, and to ensure their ads don’t appear next to extremist content. These questions are not being answered, and as a result expect to see a top 20 global advertiser pull their Facebook and/or YouTube ads for over a month in 2018.

It’s worth noting that this isn’t just a Facebook and YouTube issue. However, in Facebook’s case its continued refusal to allow meaningful third-party measurement is pretty much an own goal. For YouTube, it seems remarkable that a company that can build a self driving car can’t detect an Isis video. In 2017 (a.k.a “the year of the duopoly"), the unstoppable growth of Facebook and Google means they now are being forced to play by the grown-up rules.

We saw a glimpse of this in 2017 with a significant pull-back from YouTube. But when you look at the timeline of failure after failure from both platforms, it’s clear that advertiser reactions have been undercooked. These issues are ones that simply wouldn’t fly in any other channel.

It’s possible that the final straw may be more revelations around the US election. Advertisers may finally have had enough not because of any political meddling, but because it is only under such a strong spotlight that we find out what Facebook and YouTube haven’t yet revealed.

But put away your party-poppers duopoly haters. Because even if multiple big advertisers pull out for multiple months, you won’t see the slightest hit to revenue numbers, and just the tiniest blip in share price.

Confidence: 70%


9: Blockchain’s first killer app beyond payments and currency

While we’re heading towards the 9-year anniversary of the Bitcoin whitepaper, it’s still the beginning of the beginning for blockchains. 2017 saw a flurry of innovation (and scams) in the space. In 2018 expect to see the first useful app with over 5 million active users built on top of a blockchain.

Despite the rise of mainstream media coverage of Bitcoin, the actual innovation of blockchains is still inextricably linked to the virtual currency, and as a result usually misunderstood. Blockchains enable many things, but the most interesting is distributed computing. In 2015 Ethereum launched, combining a blockchain with a fully functioning programming language to create what founder Vitalik Buterin calls "The World Computer".

Where Bitcoin opened up the potential for basic financial apps, Ethereum opened up the potential for pretty much any app you can think of. In 2017 there was a flood of entrepreneurs doing just that, sparking the highly-scammy, probably illegal ICO boom.

If the blockchain space can get through its trough of disillusionment, there’s proper innovation coming up quickly. That innovation needs to move beyond payments and currency. But, I expect that the first killer app outside these areas to not stray too far, and will probably be built around some form of crowdfunding.

Confidence: 60% *(5m users means participants, not just speculators)


10: Commercialisation of AI will boom in the supervised learning domain

Most mainstream media around Artificial Intelligence focuses on unsupervised and reinforcement learning (or misguided chat about Artificial General Intelligence). But it’s through far less sexy applications of AI that businesses without big budget R&D teams will begin to see significant value created through this technology. Expect to see at least a couple new $1B companies emerge providing AI-based SaaS products to data-rich businesses.

Supervised learning is one of the simplest approaches to neural networks, and the technique within artificial intelligence that has seen huge leaps forward this decade. Supervised learning is great at categorising things based on lots of previous examples of those same things.

Andrew Ng (previously leader of Baidu’s AI team and Google Brain) sums it up well - anything that can be done by a human with less than 1-second of thought will soon be automated. The catch is you’ll need a lot of data to achieve this, so Ng also thinks it’s critical for businesses to have in place a data acquisition strategy to take advantage of this shift.

Confidence level: 70%


11: Regulation will cause one of the gig-economy unicorns to pull out of a whole country

If you believe the hype, the gig-economy is one of the greatest gifts Silicon Valley has bestowed upon the world. People everywhere can earn money easily, no more pesky job applications, shifts, and bosses. The only problem is that the gig deal is heavily skewed in the favour of the tech companies. So 2018 will likely see a Western country regulating, and one of the gig-economy unicorns pulling out of a whole country.

One of the defining (and powerful) features of online commerce is the ability to leverage wasted stuff. eBay let’s you sell your unwanted stuff to strangers. Facebook fills your wasted time with baby photos. AirBnB let’s you monetise your wasted spare room.

Mobile levelled-up this trend. If you own a car and a mobile phone, you can be an Uber driver. No car? Grab a bike and deliver some food for Deliveroo.

These things look a lot like jobs. But they’re not - just ask the startups! Your drivers and riders (and hosts) usually don’t get insured by their employer, they have no rights to minimum wages, and they definitely don’t get holidays or super. In Australia your Uber driver has to pay the GST for your trip out of their own pocket.

2017 saw many cities start to question this arrangement - most notably Austin and London. The easy retort is that powerful taxi lobbies in the like are pulling the strings. And while that may be true, there is undeniably something not quite right about these gig-tech unicorns.

Confidence level: 80%


12: The world’s biggest ad fraud network uncovered

There’s not a lot of people banging the drum about ad fraud, but it continues to be a huge problem. Like $13 billion huge. I don’t expect marketers to care a lot more about ad fraud in 2018, but I do expect the largest ever ad fraud network to be revealed as being responsible for stealing more than $100m of ad budgets.

Ad fraud isn’t really just one thing. It’s a bunch of tactics that are constantly evolving. It’s also often quite technical, so it makes sense that the marketing blogs and trade news pretty much ignore the problem (interestingly, domain spoofing is relatively easy to explain, and seems to have gained a decent amount of press this year).

What I find amazing is that ad-blocking regularly causes mass-hysteria with agencies and marketers. Yet ad-blocking doesn’t actually harm advertisers, it just makes audiences harder to reach.

But ad fraud does cost advertisers. The result of ad fraud is that organised criminal gangs end up with your marketing budget. It baffles me that more people don’t care about this.

As long as ad fraud remains the most profitable way to steal money on the internet (arguably only challenged by ransomware), ad fraud will continue to advance. As it becomes more advanced, it will scale to phenomenal new heights. And still it’s unlikely many will care.

Confidence level: 80%

Postscript: A few days after this was written, researchers uncovered “Hyphbot”. An ad fraud network running since August that may have been generating around $500,000 per day.


- November 2017