What’s in the pipeline vs what’s been projected: improving the accuracy of energy scenarios

Excerpt from Clean Energy Action research, 5 Oct 2016

The projections published in the EIA’s Annual Energy Outlook (AEO) have invariably overestimated the cost of renewable electricity generation and fallen sadly short of predicting new additions of wind and solar capacity. For example, Figure 1 shows that the projections published in the EIA’s Annual Energy Outlook repeatedly underestimated U.S. utility-scale solar photovoltaic (PV) capacity from 2011 to 2015 and continue to predict that solar installations will largely stall through about 2025.

In reality, however, solar PV capacity is growing at an unprecedented rate. The Solar Energy Industries Association reported that by the third quarter of 2016, the cumulative U.S. utility-scale solar PV capacity (including capacity which was under contract but not yet operating) exceeded the AEO2015 projection for capacity in 2039. Accounting for planned capacity which had been announced but was not yet under contract by Q3 2016 indicates that utility-scale solar PV capacity will soon far surpass all AEO projections for 2040.

Solar PV Capacity and Projections
EIA reference case projections of U.S. utility-scale solar PV capacity and historical data (black, bold) as well as points which include planned capacity under contract in Q3 of 2016 and announced but pre-contract installations as of Q3 2016. Projection data taken from the EIA’s Annual Energy Outlook, historical data taken from Solar Energy Industries Association’s U.S. Solar Market Insight Reports.

In addition to missing the sharp rise in solar photovoltaic installations, EIA projections also missed a dramatic downturn in coal production over the last decade. They failed to pick up on the trend year after year and still predict flat or rising coal production through 2040, as shown in Figure 2.

History (black, bold) and annual EIA projections of U.S. coal production from 1997 to 2040. Note that the vertical axis starts at 950 million short tons for clarity. Data taken from: the EIA's Annual Energy Outlook.
History (black, bold) and annual EIA projections of U.S. coal production from 2006-2015. Note that the vertical axis starts at 950 million short tons for clarity. Data taken from: the EIA’s Annual Energy Outlook.

Disruptive innovations tend to precipitate new market trends that are notoriously difficult to predict. Just as the invention of the personal computer led to an abrupt decline in the typewriter industry in the late 1900’s, a massive transition toward renewable resources is transforming U.S. energy markets and so far EIA projections have failed to keep up with this transition. Every year, EIA forecasts predict a return to the trends of the 90’s, but the technological and political landscapes surrounding the U.S. energy industry are changing rapidly and historical precedent suggests that energy markets may never return to those of past decades.

For more details, readers are encouraged to download the full CEA White Paper here.

As Jeremy Leggett says (Feb 2017):

First and second, climate and energy:

More than a thousand cities are committed to 100% renewable power. So are more than 80 of the world’s biggest companies, in Google’s case as soon as this year. More than 600 financial institutions worth more than $5 trillion are pulling their capital out of fossil fuels.  Other investors are pressuring fossil fuel companies to invest more in clean energy, and their numbers are set to grow. The G20 Financial Stability Board’s Taskforce on Climate Related Financial Disclosure involves companies with market capitalisation of $1.5 trillion and financial institutions responsible for assets of $20 trillion. It aims to build a market in transition to a sub-two-degree world.

A global energy transition from fossil fuels to clean energy is unfolding before our eyes, and fast, and not just because of serious intent in climate action. Solar and wind power will be the cheapest option in most countries within just a few years, and in some sectors and countries already are. Cheap batteries will soon be storing their electricity on a massive scale. Electric vehicles are on course to knock an entire category of oil use, diesel, out of the markets within ten years.

Meanwhile, as fast as the prospects for clean energy rise, those for fossil fuels fall. China has just cancelled more than 100 coal power plants, dozens of them already under construction. India says it needs no more coal in the next five years. The oil industry is struggling under a multi-trillion dollar collective debt mountain, with analysts increasingly giving up hope of long term profitability, even if oil prices rise. Investors unconcerned by climate change are abandoning them, and oil-rich Gulf states are drawing up plans for an an end to oil within five decades.

Trump may want to dig coal, but how can he do so when it is uneconomic compared to solar and wind? One in 50 US jobs is now in the solar industry, and the growth rate is accelerating: up a record 25% in 2016. Is he going to axe those jobs somehow? When 73% of his own voters support more solar and wind, according to a recent poll?

Accordingly, I set the dial on energy as distinctly positive. I summarise many more reasons for doing so in my blogs spanning 2016 and my book The Winning of The Carbon War.

Third, tech. Turning this dial up will require artificial intelligence and robotics to be applied with appreciable net benefits for society as a whole.

The development of AI and robotics is evolving even faster than clean energy. On one hand, profound social benefits are in prospect. Medical diagnosis is a good example. Computers machine-learning while using global databases are providing life-saving diagnoses that elude human medical experts. They are making it less difficult to hold criminals to account. The UK’s Serious Fraud Office used a crime-solving robot to plough quickly through the libraries of information necessary to prove systemic corruption in Rolls Royce. Control of electricity demand by AI in data centres is achieving remarkable emissions cuts in what is a large global point source of carbon dioxide. Such examples are plentiful.

But on the other hand tech leaders are openly worrying about the effect of exponential AI and robotics on jobs. In Japan, the government is looking to robotics to boost the national economy, and robots already outnumber humans in the kitchen of one restaurant. In the UK, banks are preparing to roll out robot tellers aiming to improve customer service via machine-learned empathic responses. In the US, the world’s largest hedge fund is working to replace managers with AI, including in HR and strategy roles. Wired, the magazine of choice for many in the digital world, concludes as follows: “The AI threat isn’t skynet. It’s the end of the middle class.”

Concern over job losses is fuelling a significant component of the anger expressed by the populist right. Meanwhile, the potential downsides of AI and robotics when it comes to authoritarian regimes do not need much imagining. In the wrong hands, uncontrolled, they can quickly amount to the perfect infrastructure for police states.

Thankfully, these dangers are being recognised by both scholars and practitioners. In 2016, Prof Steven Hawking and other luminaries published a letter pledging to ensure that AI research benefits humankind. In January 2017, hundreds of AI and robotics researchers developed 21 principles – the Asilomar Principles – for use of the new technologies. Principle 23 involves Common Good, and reads: “Superintelligence should only be developed in the service of widely shared ethical ideals, and for the benefit of all humanity rather than one state or organization.”

To ensure that, a global convention on the use of AI would seem like a good idea. It would offer an opportunity to build on the renewed confidence in multilateralism engendered by the Paris Agreement. The Asilomar Principles would be a good starting point for a draft. The fact is that we are wiring the world way faster than any consideration of public policy, legal, ethical, and human rights implications, and doing so just at the time demagogues are on the rise. Even mainstream analysts are pointing out that the demagogue currently occupying the White House could easily morph into a despot, and his early attitude to the US judiciary and press are more than consistent with this.

All this being the case, the net global score for the tech dial should probably be set slightly negative. This assessment excludes the role AI plays in the next theme. Including that would drive the score much higher. 

Fourth, truth. Turning this dial up will require tech to be used for improving the processes of liberal democracy, including quality and verifiability of information, and the transparency thereof.

The rise of demagogues has been much assisted by the explosion in 2016 and 2017 of so-called fake news, otherwise fairly describable as systemic lying in mass media. Famous examples of the falsehoods pushed on populations include Pope Francis supporting Donald Trump, and the UK Brexiteers’ insistence, contradicted by the UK Statistics Authority among others, that Britons “send the EU £350 million a week” that could be spent instead on the NHS. Analysis of both the UK Brexit vote and the US Presidential election shows how pervasive the problem has become. As Wired put it, in 2016 “the Mainstream Media melted down as fake news festered.” By August fake news about the US election was increasingly outperforming the top stories at the 19 major US news outlets.

One of the best forensic analyses of how this happens is by journalists at Scout. They describe four ways in which AI has been “weaponised” by organisations owned by conservative and alt-right interests, such as Cambridge Analytica, and leaders of the far right such as President Trump’s chief strategy advisor, Steve Bannon.

The first involves the marriage of big data surveillance and computational psychology. Cambridge Analytica has created a model on this basis that compiles individual predictive personality profiles for entire populations. Predicting a subject’s behaviour better than their partner could requires analysis of just 300 Facebook likes, so one expert professes. This before considering many other easily available analyses of online behaviour, and numerous sources of megadata that can be bought.

The second involves automated engagement scripts that prey on human emotions. The practitioners call this “behavioural micro-targeting” and it allows them to change behaviour on an industrial scale. In the US election, personally targeted ads were used extensively in key swing areas. These included Facebook “dark posts”, visible only to those targeted. Cambridge Analytica were able to monitor peoples’ responses, such as sharing on Facebook, and assess which messages were resonating, tuning advertising accordingly.

The third involves creation of propaganda networks that rapidly accelerate ideas. The purveyors of fake news have spent years building websites, and these now operate on a scale that allows gaming of search engine optimisation, including crucially at Google.

The fourth involves use of an army of bots – computer programmes that talk like humans on fake social media accounts – to echo fake news and police public debate, including by intimidating and suppressing the voice of the opposition.

The problem is going international, fast. Brexit-backer and Trump ally Arron Banks plans to launch a “news” site in the UK that aims to repeat the successes of US lie machines like Breitbart, the news agency founded by Steve Bannon. It will be pro-Brexit, pro-Farage, pro-Trump, anti-establishment, anti-open borders, and anti-corporatism. In the US, where rightist billionaires outnumber those openly supportive of liberal causes, figures like fossil-fuel tycoon David Koch are funding campaigns to smear prominent potential champions of counter views like Elon Musk.

Meanwhile, it is not as though rightist populism goes unfueled in the editorial suites of the mainstream media. In the UK tabloid press, it dominates. When judges ruled in November 2016 that Parliament should vote on Brexit, the Daily Mail published a front page headline “Enemies of The People”. Believers in functional democracy attack their independent judiciary at their peril.

This is all happening today. Tomorrow? Silicon Valley guru Peter Diamandis warns that within four years – by the time of the next US election – AI will be ten times more powerful, and will be being applied to 50 billion devices and a trillion sensors. Digital avatars will be photorealistic and fully programmable, able to have instant conversations with citizens. Picture standing in a taxi queue having a conversation with an avatar of a candidate on a screen. The operators of the sensors in the screen, having auto-identified you from facial recognition databases, will know from your profile if you are susceptible to any of their candidate’s arguments. They will know more about you than you do yourself. The candidate’s avatar will ask you a leading question, targeting your emotions. You respond. The candidate responds back in real time, with libraries worth of machine learning picking the answer used.

All this manipulation will be unfolding in a world becoming inexorably more permissive of mass surveillance.  In November, the UK legalised – barely noticed – the most extreme surveillance powers in the history of western democracy. As Ed Snowden commented at the time, the Investigatory Powers Act goes further than many autocracies. It was passed despite the fact that only a few weeks earlier the Investigatory Powers Tribunal had ruled that UK security agencies have been illegally collecting data for 17 years, in violation of the European Convention on Human Rights.

So much of this drama focusses on the creations of Silicon Valley, where company founders and employees have tended to favour the Democrats over the Republicans. What is the response there?

The pushback has begun, but it hardly amounts to a resistance consistent with the scale of the problem yet. Over 100 Silicon Valley firms have filed a legal brief opposing President Trump’s executive order banning travel from seven mostly Muslim countries. These include Apple, Facebook, Google, Microsoft and Tesla. Employees of tech firms rallied in their thousands against the ban. Customers projected an intense displeasure in the marketplace: tweets protesting Uber’s links to Trump pushed competitor Lyft ahead in downloads for the first time. Uber’s CEO felt the pressure so strongly that he resigned from Trump’s economic advisory council. But none of this has addressed the process by which Silicon Valley’s innovation with megadata is being hijacked by the far right yet.

Facebook founder Mark Zuckerberg is aware of the problem. In a 5,700 word Facebook post, he has endeavoured to chart a responsive course for his creation. It was met with skepticism. “Mark Zuckerberg’s Answer to a World Divided by Facebook Is More Facebook”, Wired concluded. Guardian tech correspondent Alex Hern, wrote a withering paragraph-by-paragraph dissection of it. Nikki Usher, Professor of new media and technology at George Washington University, concluded: “He might be in denial, because a lot of the rest of us are.”

All this considered, the global score for the truth dial, as things stand, must surely be set at net severe negative. Already. Far worse is eminently conceivable, in relatively short order, unless resistance can be marshalled effectively.

Fifth, inequality. Turning this dial up will require significant narrowing of the income gap, both within the developed and developing worlds.

Many analyses of the rise of the new demagogues show that rage over a widening income gap strongly influences those prepared to vote for them.

  • The figures are shocking in America. Between 1970 and 2014, average income grew 77%, but almost all of these gains went to the top 1% of earners.
  • Elsewhere, elites have also allowed disproportionate self enrichment to run rife.
  • Globally, the 8 richest people own the same wealth as the poorest 50% and the richest 1% own more than the other 99%.
  • Every year at the World Economic Forum, attendees openly worry about the unsustainability of these figures, and the social divisiveness they create. Yet each year they do precious little about it.

Efforts to reduce inequality have seen some limited success at the bottom of the wealth league table. The percentage of people in extreme poverty – those earning under $1.90 a day – is falling in all regions of the world. But it must fall far faster if the World Bank is to hit its target of eliminating extreme poverty by 2030.

The global score for the inequality dial as things stand, must be set at a clear net negative.  Ed. note:  poverty levels set at higher thresholds than the World Bank’s $1.90/day continue to rise.  The WB’s measure seems to obscure this issue.

 Sixth, reform. Turning this dial up will require much more attention to market failures.

 The need for significant reform of capitalism has been widely acknowledged since the financial crisis of 2008. The absence of it has been the subject of rage even in the conservative press. One Daily Mail headline in June 2012 exhorted “Put Bankers In The Dock”. The virtual absence of legal redress for financial malfeasance, even in the case of companies and executives with hands caught jammed in the till, has undoubtedly contributed to the rise of the populist right and the demagogues they support.

In the US, the Dodd-Frank Act went some way to introducing checks and balances on Wall Street. Having railed against the excesses of bankers in his campaigning, Trump in office wants to roll back even that.

With the advent of the VW “dieselgate” scandal, and the prospect of multiple executive jailings for the systemic fraud involved, it may be possible that the tide could beginning to turn. But many of the proposals put forward for re-engineering of modern capitalism in the wake of the financial crisis, even by pillars of the capital markets, have gone by the wayside. Bonuses in the US and UK today are even bigger, in real terms, than they were pre financial crisis. World debt has been allowed to balloon to more than $150 trillion, a record. Out of control bonuses and mountainous unsecured debt were two of the main triggers of the 2008 global financial crisis.

The global score for the reform dial, as things stand, must accordingly be set at clear net negative.

Seventh, conflict. Turning this dial up will require a proliferation of common security in the world

We live in a world where superpowers are fighting wars by proxy. Russia and Syria stand accused of deliberately “weaponising the refugee crisis” – as one NATO official has put it – in an effort to destabilise Europe. Meanwhile the superpowers are probing each others’ cybersecurity with increasingly sophisticated malware. Pointers to the pervasiveness of this new form of conflict bubble to the fore from time to time. In January 2016, hackers shut down the Ukrainian power grid. Kiev accused Russian special forces. The malware involved had previously infected power suppliers in the US and Europe, without shutting down supply. It is widely suspected that malware sits waiting to be triggered throughout infrastructure in the superpowers. Given the fragility of US electricity grid infrastructure this should be a particular concern. The three US grids are aged, with large power transformers on average 40 years old. The US suffers more blackouts than any other developed nation.  Centcom Commander General Lloyd Austin has said of the grids’ susceptibility to attack: “it’s not a question of if, it’s a question of when.”

The CIA, NSA and FBI have all concluded that Russia tried to influence the US presidential election in an effort to get Trump elected. We can only imagine what they know of Russian hacking of trojan horses into the US power grid.  For their part the Russians have to fear American capabilities too. An attack on Russian bank Sberbank in late 2014, for example, hints at the vulnerability of their capital markets. It spooked depositors into withdrawing $20bn in one week.

Other states would appear to be playing the same kinds of games. Saudi Arabia has blamed Iran for serious cyber attacks on its aviation authority and on four other unnamed bodies.

Dangerously destabilising as these proxy conflicts are, the potential for cyberattack on the world’s nuclear weapons, and their aged software support, hardly bears thinking about. The reduction in global warhead inventory from around 70,000 in the mid 1980s to some 15,000 today has been a somewhat positive feature of the years since the cold war. Yet both Putin and Trump have recently said they want to “strengthen” their nuclear weapons stockpiles.

The global score for the conflict dial, as things stand, must be set at a manifest net negative.

This completes my survey of the state of play on 20th February 2017. I will continue to review the evolving drama periodically from here on. In a subsequent blog, I will examine options for an integrated global plan aiming to tip all the seven dials into net positive territory, en route to an appropriate civilization, and away from the new despotism. I need hardly add that the challenges involved in so doing will be without precedent. But there is nothing in the sum of human affairs that is more important. — Jeremy Leggett