The science of global warming is unclear:

• The historical temparature record is debatable.
• The extent of the human contribution is unknown.
• Projections of mean temperature increases vary widely.
• Estimates of how that change will be distributed between currently warm and cold regions, and of its impact on, e.g., precipitation patterns, are even less reliable.
• Current models incorporate lots of positive feedbacks, such as reduced reflectance and thus greater heat retention from shrinking the polar ice caps. But they may miss negative- feedback processes that would moderate any warming effect.
• Assuming that warming is real and partly anthropogenic, there’s no assurance that reducing emissions now will stop the trend, which may be self-reinforcing once started.

2. Even if we knew the climate science, that still wouldn’t tell us about the impacts on humans and other species:

• The results will be favorable in some places and unfavorable in others.
• The mix is unknown.
• Humans and their societies adapt to change by changing their behavior. Current models of the damage done by global warming take inadequate account of human ingenuity in adjusting to change, especially when that ingenuity is mediated by the price system. Price changes tend to increase the production and decrease the use of scarce resources. Technologies can change very rapidly under the pressure of changing prices.
• Other species adapt to change, or fail to, in a constant Darwinian process. It’s not obvious that increasing the rate of change will have on balance bad rather than good impacts.

3. Assuming that global warming is happening, that one of its causes is human activity and in particular the release of greenhouse gasses, and that its net impacts are bad, we still don’t know that major interventions now to control global warming will do more good than harm.

• Money spent on controls can’t be spent on other things.
• If the money that would be required to implement strict controls on greenhouse-gas emissions were instead expended on, e.g., providing clean drinking water and female literacy for very poor countries in Africa and Asia, the suffering averted would far outweigh the benefits of having a somewhat cooler planet a century from now.
• In particular, if the citizens and governments of developing countries were given a choice between spending money now to avoid the impact on those countries of global warming in the future, and spending money now to relieve current poverty and increase the rate of economic growth, they’d choose development assistance over climate control.
• The richer and more technologically advance the world gets, the better it can adapt to whatever climate change occurs, as well as to more routine events such as hurricanes and tsunamis. Reducing the rate of economic growth will tend to reduce the rate of technological innovation, thus potentially retarding the development of better means to control global warming or to mitigate its impacts. So imposing expensive controls now may be counter-productive.
• Controls may have unpredicted and undesirable side-effects.
• Most of the predicted impacts occur in the far future. That makes taking current action to prevent those future impacts a dubious proposition. It’s rational to discount the far future compared to the near future because resources not spent now can be invested, creating future value. That applies to building factories, making scientific discoveries, and educating children.
• World standards of living have been rising sharply for the past 200 years, and there is no evidence that the growth in living standards is slowing down. Imposing costs on the current population of the planet to create benefits for far-future populations means transferring income from the poor to the rich.

4. There may be technological fixes much cheaper than reducing greenhouse-gas emissions. for example, increasing the albedo (reflectance) of the Earth by putting light-reflecting particles in the stratosphere. Small investments now in developing such fixes will have greater returns per dollar than large investments in reducing emissions.

5. Even if we wanted to take action now, existing international institutions will not support an effective greenhouse-gas-emissions-control regime, and an ineffective one, such as proposed in the Kyoto Protocol, would generate large costs and small benefits.

6. Therefore, we should pursue a "no-regrets" policy: making investments now that make sense on other grounds – e.g., developing alternative fuel systems for automobiles – and that will also pay off in terms of controlling climate change if that turns out to be a problem, and studying better means of both controlling warming and understanding and adjusting to its effects, but not making huge sacrifices now to prevent we know not what.