You've probably heard that US carbon emissions have been falling. According to President Obama and energy commentators the world over, fracked shale gas has displaced dirty coal, in much the same way that fossil fuels undercut whale oil a century earlier. Out with environmentally unfriendly old technologies and in with cleaner and more efficient new ones. Everyone wins – including the climate, thanks to the fact that gas produces only around half as much CO2 as coal does for each unit of power or heat created.

On the other hand, you may also have heard that US coal exports have increased as its domestic emissions have fallen. America currently has little in the way of gas export facilities but plenty of capacity for shipping coal to Asia, Europe and elsewhere. Those ports have been busy of late and the ripple effects are being felt far and wide. For instance, UK emissions shot up 4.5% last year, partly due to low coal prices made possible by surging US exports. So could it be that rising US gas production has increased the human contribution to global warming, even as American's own emissions have fallen?

This is an important question. There are legitimate and complex debates to be had about potential of natural gas as a 'transition fuel'; and about how gas's huge benefits over coal in terms of carbon emissions and air pollution should be weighed against the methane leaks and water pollution linked to fracking. But in the absence of a global emissions deal, those debates are largely academic unless producing more gas does actually lead to less coal use. So does it?

To explore this issue I propose considering an unfamiliar metric: not carbon emissions but carbon extraction. After all, the climate doesn't care where a unit of carbon is burned; just whether it comes out of the ground and enters the atmosphere. It makes sense to ask, therefore, whether the US is now extracting more or less carbon than it was before the shale gas boom. Has coal production fallen enough to offset rising gas production?

I've just had a go at working out the answer to this question by taking government data for US oil, coal and gas production and adding up the carbon that each fuel contains. The results are interesting. (My dataset, conversions and sources are here. Error-checking much appreciated. Tweet me on @theduncanclark if you spot anything awry.)

First, let's look at the three fossil fuels individually. The graph below shows the carbon content of US-produced natural gas rising quickly since around 2005 as fracking rigs multiply, just as you'd expect. The same is true for oil, which had been on the wane since American crude production peaked in the early 1970s but is now rising again thanks to new fields and production techniques, bucking the bell-curve trend predicted by peak oil theory (for now at least). By contrast, coal extraction fell sharply after the 2008 financial crash but has been creeping back up since.

Now let's stack all those fuels on top of each other to see total carbon extraction. The resulting graph shows that that there has been no decline in the amount of carbon the US is taking out of the ground. In fact, the trend is upwards. The latest year for which full data is available – 2011 – is the highest level on record, at 5288 MT of potential CO2, a fraction higher that the previous peak in 2008. Despite or because of the shale gas revolution, 'American carbon' is flowing into the global economy and atmosphere faster than ever.

Of course, this ignores any imported fuels that the US has managed not to burn by producing more of its own oil and gas. But given that global emissions keep rocketing upwards, it seems likely that all of those fuels have simply been burned elsewhere, just like the unneeded US coal.

It's possible that things may look a little rosier next year. The provisional data point to a dip in US coal production in 2012 and exports are down this year due to a glut in Asian supply. The recent announcements by the World Bank and the European Investment Bank may depress coal investments too, as will Obama's promised new rules on power plant emissions. On the other hand, US oil and gas production are still on the up and the government has predicted a mammoth increase in global coal consumption in the next few decades. But whatever happens in the future, the clear conclusion so far is that while gas has helped reduce US carbon emissions, it hasn't reduced US carbon extraction.

An age-old feedback loop

None of this is surprising when you consider the history of human energy use. At the global level no new major energy source has stopped the existing sources from expanding. Indeed, as Mike Berners-Lee and I argue in The Burning Question, despite radical changes in the global energy mix over the last two centuries (and even more radical changes within individual countries) energy use and carbon emissions have undergone remarkably consistent long-term exponential growth. The implication is that there's a technological and social feedback loop at work, with each new energy source increasing access to and demand for all the other sources. Energy begets energy.

The graph below, which shows total human energy use since 1850, reflects this. When coal use took off in the nineteenth century biomass energy didn't decline as is often assumed. In fact it increased, helped rather than hindered by coal-powered industrialisation and globalisation. Similarly, coal use increased when society started extracting large amounts of oil – which makes sense given that oil not only proved useful for coal mining but also enabled the mass roll-out both of cars and energy-hungry suburban homes. In turn, gas and hydro helped drive technological and engineering revolutions that have made obscure oil sources more viable.

(Whale oil – often cited in the shale gas debate – is one exception to this rule, but it was never a really substantial energy source. Capturing energy with a harpoon was always difficult, expensive and dangerous, so it's no surprise that the use of blubber-based fuels collapsed quickly when alternatives came along. That said, sperm oil was still used as lubricant in gearboxes until the 1970s when it was banned for conservation reasons, much to the annoyance of General Motors, which blamed the phase out for an embarrassing run of faulty cars.)

The fact that new energy sources tend to be additional to existing ones helps explain why more gas production has dinted neither US carbon extraction nor global emissions. But critics of gas beware: the same caveat applies to genuinely low-carbon energy sources such as renewables and nuclear, or indeed increasing energy efficiency. We usually assume that installing a wind turbine or nuclear plant will reduce global emissions but that's not necessarily true, since the fossil fuel that the clean energy system replaces may get burned elsewhere instead, perhaps kick-starting new energy feedback loops in other parts of the world and driving global carbon emissions up yet further.

In some cases there has even been talk of using low-carbon energy sources directly to increase fossil fuel flows. For example, modular nuclear reactors are being considered as a way to propel natural gas down the remote pipelines that bring energy to Europe's homes and power plants, or for melting tar to produce oil for the world's billion-strong car fleet. This seems crazy at first given that it would be more efficient and less polluting to use the nukes directly for producing electricity, but existing infrastructure can determine our energy choices as much as the available energy sources do.

Whether or not we end up using low-carbon power to produce more fossil fuels, the graph above should make us cautious of the assumption that either gas or carbon-free energy sources will in themselves reduce emissions. Yes, in the very long run fossil fuels will become uneconomic compared to the alternatives, no matter what happens with climate legislation. Yes, we need massive effort and investment in clean-energy R&D to bring that date forward. And yes, with a global carbon cap in place, if one can ever be achieved, more gas would have to mean less coal, which would bring huge benefits in terms of climate change and local air pollution. But there's little evidence so far that fracking, wind power, nuclear or any other technology is helping us leave any carbon in the ground. Indeed, as I wrote recently, despite all the renewable power installed so far, all the fracking rigs, all the energy efficiency gains, all the national carbon cuts, and even a collapse in average fertility levels, global emissions are still growing at the same rate today as they were in the 1850s.

This context is why, in my view, people concerned about global warming should spend less time arguing over the relative merits of renewable, nuclear and gas and more time focusing on measures that directly stem the flow of carbon out of the ground and into the air. Cleaner power sources and increased efficiency are crucial but they won't be enough if in parallel all nations continue to rip fossil fuels out of the ground as fast as possible.

Whose carbon is it anyway?

One question all this raises is which nation is responsible for each tonne of carbon. If, for instance, as a result of its gas boom America exports some coal to China where it gets used to produce T-shirts for UK consumers, who is to blame for the emissions that result? Many green thinkers have argued implicitly or explicitly that the end consumer – British shoppers in this example – bear full responsibility. But that's a very black and white view of the world. Does a teenager spending their Saturday job wages on a new T-shirt really bear more responsibility for that carbon than a US coal company which produced it and may be lobbying Congress to support new mines and export terminals? Of course not. Does China have at least some responsibility for the design of the energy infrastructure it uses to supply the world with manufactured goods? Surely yes. In other words, responsibility for those emissions is shared between everyone involved in the carbon supply chain, from extractor and manufacturer to end consumer.

While it wouldn't be fair to hold the US wholly responsible for the emissions that result from its exported coal, therefore, those exports do seriously undermine claims that the nation is making real progress on global warming. Even if America's emissions don't rise again (which they might as the economy grows, especially if more gas gets exported and more coal gets used domestically) this inherently global problem can't be solved by pushing the carbon around between nations to avoid leaving it in the ground. With every shipment of coal from the US to the rest of the world, the cost of dirty energy goes down, global emissions go up and an agreement to phase out carbon fast enough to avoid dangerous climate change becomes less likely.

Does President Obama understand this? I suspect he does. And yet as David Roberts pointed out on Grist, when Obama finally decided to take a stand on climate change a couple of weeks ago, his long speech was conspicuously quiet on US coal production. He did promise to tighten up power plant emissions standards in the US – a significant step forward. But there was nothing on mining in Powder River Basin or the proposed coal export terminals in Oregon and Washington.

Perhaps that shouldn't be surprising. It's politically painless to talk up the benefits of gas, renewables and efficiency, even if global emissions keep accelerating upwards at the same rate as before. It's much harder and riskier to do what's actually required: consciously wind down coal production as part of a global effort to slow the extraction and use of fossil fuels. Obama is already accused by his critics of waging a 'war on coal' and so far he's hardly even got his hands dirty. If he went after coal in any serious way he'd have a major political fight on his hands, as anyone who grew up in Britain in the 1980s can testify. (The context and political motivations for Margaret Thatcher's dismantling of British coal mining were completely different, of course, but the example is pertinent nonetheless.)

Ultimately, however, if we want to avoid frying the climate, the world needs to make some hard choices about fossil fuel production – and especially coal mining. Even if we get lucky and the climate turns out to be less sensitive to CO2 than earlier estimates (which is not by any means a given), global emissions still need to be falling quickly to give a decent chance of limiting global warming to 2C – the agreed global target that looks both less plausible and more risky with each year that goes by.

If we can't bring ourselves to start constraining fossil fuel flows soon, we'd better be going hell for leather developing not just alternative energy sources but all forms of carbon capture technology, from traditional CCS to carbon negative materials and farming. Because if we want to keep the planet safe, we either need to leave most of the remaining fossil fuel in the ground undisturbed, or burn it and capture the carbon. There is, unfortunately, no third option.

Duncan Clark is coauthor of The Burning Question: We can't burn half the world's oil, coal and gas, so how do we quit?