I think the sentence does make sense, though it is certainly badly written!
Thanks for this vigorous and informed replies!
1. You say that IAM’s don’t factor in economic factors. I think this is wrong, or perhaps I have misunderstood your point? IAMs model the role of different energy technologies in an energy system meeting an emissions and economic constraint. The typical IAM does indeed imply a quadrupling of nuclear to 2050 (Peters et al, p4). This suggests that you are wrong to give the impression that all experts believe that nuclear should be phased out. As another example, the authors of the Clack et al response to Jacobson et al are all highly respected energy researchers who believe that at least 20% of energy needs to come from firm controllable low carbon sources. This means either gas+ ccs or nuclear, right?
2. It was very difficult until recently for private Gen IV nuclear companies to operate due to licensing and regulation. This was the point I was making.
Korea and UAE also show low cost and labour costs there will be comparable to the US/EU.
a. System levelised cost is important because it tells us what technologies we will need in a completely decarbonised system. It is at the system level where the case for nuclear becomes clear.
Re betting—my point was that there appear to be lots of potential barriers to VRE, including uncertain cost, local opposition, system grid balancing etc. There are a range of studies which suggest that with a zero emissions constraint, costs increase nonlinearly as VRE penetration passes 50%. Do you deny this?
i’m surprised you give the example of batteries to make your case here. A VRE-dependent system has *multi-week* electricity droughts to which batteries are ill-suited. It might be true that we get long duration storage but the technology isn’t there yet.
b. The problem I was highlighting was of VRE *relative to* nuclear. VRE at high penetration has colossal land use requirements. David McKay argued that due to local opposition, VRE could at best provide one sixth of electricity in the UK. Nuclear in contrast because it is very energy dense doesn’t have the same concerns.
4. I think you are missing my point on Germany. Germany has indeed subsidised VRE a lot and this has indeed helped to bring costs down. However, the point I was making was that Germany’s own domestic performance on climate targets is bad relative to Europe, but they are following the exact strategy you propose—no nuclear, lots of VRE. Onshore wind additions fell to a twenty year low last year in part due to local opposition—they are around a quarter of what they need to be for them to get to their target if 65% of electricity from VRE. I think it highly likely that this local opposition to lots of wind will be replicated everywhere else where it is proposed.
5. The fastest rates of decarbonisation in absolute terms are from VRE. I don’t think this is right, but perhaps I am not understanding what you mean. France and Sweden nearly completely decarbonised their electricity supply in ten years with nuclear/nuclear+hydro. I don’t think anyone is on course to do that with VRE are they?
6. On zero carbon fuels, the problem here is that VRE would mean that the electrolyser is not used most of the time, whereas nuclear could run the electrolyser at max capacity. From memory the electrolyser is a third of the cost of hydrogen production. From modelling I’ve seen, VRE would be a very expensive way to produce liquid fuels.
It may be noted that in the thing I wrote on climate change I don’t actually defend long-termism or even avow belief in it.
For those who find it confusing that I, at best a mid-table figure in EA, get dragged into this stuff, the reason is that I once publicly criticised a post on Pinker that Phil wrote on Facebook (my critique was about three sentences). Phil has since then borne a baffling and persistent grudge against me, including persistently sending me messages on Facebook, name-checking me while making some rape allegations against some famous person I have never heard of, and then calling me a white supremacist. Hopefully, this gives some insight into Phil’s psychology and what is actually driving posts such as the one linked to here.
GiveWell seems to be unusually well run.
What are your top organisation, management and strategy tips? Research management tips would be especially useful
Allocating time to research is a difficult optimal stopping problem—how do your researchers decide when they should stop researching a particular question or subquestion?
How likely do you think we would be to recover from a catastrophe killing 50%/90%/99% of the world population respectively?
What are your top three productivity tips?
What is your solution to Pascal’s Mugging?
Do you think the problems of infinite ethics give us reason to reject totalism or long-termism? If so, what is the alternative?
Do you think we will ever have a unified and satisfying theory of how to respond to moral uncertainty, given the huge structural and substantive differences between apparently plausible moral theories? Will MacAskill’s thesis is one of the best treatments of this problem, and it seems like it would be hard to build an account of how one ought to respond to e.g. Rawlsianism, totalism, libertarianism, person-affecting views, absolutist rights-based theories, and so on, across most choice situations.
Is your view that:
(i) the main thing that matters for the long-term is whether we get to the stars
(ii) This could plausibly happen in the next few centuries
(iii) therefore the main long-termist relevance of our actions is whether we survive the next few centuries and can make it to the stars?
Or do you put some weight on the view that long-term human and post-human flourishing on Earth could also account for >1% of the total plausible potential of our actions?
Does it worry you that there are very few published peer reviewed treatments of why AGI risk should be taken seriously that are relevant to current machine learning technology?
What do you make of Lant Pritchett’s arguments against the RCT-based approach to development, and for focusing on national development?
This feels more like something that could be conveyed on 1 side of A4. Could someone create a webpage with the evidence on CBT-i and instructions on how to do it?
1. I think you give a partial picture of the split in expert opinion here in the penultimate paragraph. I think it would be more accurate to say that some people take the view you do and some respectable people take the view that firm controllable low carbon power will be very important. e.g. Your headline claim is pretty strongly at odds with IPCC integrated assessment models, which the typical model saying that a quadrupling of nuclear is needed, rather than the controlled mothballing that you suggest here. And these models also assume a massive increase in bioenergy with CCS, which seems very unlikely to happen, suggesting that nuclear will have to step in.
2. The picture you give on cost ignores where most nuclear new build is happening today. The vast majority of new nuclear is built in China at the moment, and the typical plant construction time is around 6 years, with costs at around $3000/kW. This shows that failures in the US and Europe are particular to the politics and licensing regime and to the industry, rather inherent to the technology. And it shows that changing the licensing regime to allow next gen nuclear in the US and Europe could make a large difference.
3. It is useful to think about the role of nuclear as one about reducing the risk of our decarbonisation efforts. On your approach, I take it that we would bet on solar and wind continuing to get cheaper and then taking over 80% of electricity. To me, it is much safer to invest in the full range of low carbon tech options, including nuclear, if there turn out to be barriers to getting to 80% solar and wind.
Technology-level levelised cost is a meaningless metric. The more relevant one is the system-level levelised cost. Studies show that once solar and wind go past 50-80% of electricity production, system-level costs start to rise dramatically. Should we bet on the inflection point in the real world being 80% rather than 50% or 40%? I would prefer not to.
Cost is only one determinant of political feasibility. The studies you mention I take it refer to increasing long-distance transmission infrastructure 2-4x. The land use requirements of this and of high solar and wind systems are enormous. There is already significant local opposition to onshore wind in the UK, where it supplies about 9% of UK electricity—there was a ban on subsidies until recently even at these levels of penetration. In Germany, new onshore wind has flatlined in part due to local opposition.
Value deflation at higher levels seems to be a major problem in high-solar and wind systems. Various studies suggest that for example solar would have to decline well in advance of this historical trend in order to outpace value deflation. https://www.vox.com/2016/4/18/11415510/solar-power-costs-innovation
The route you propose is the one Germany has taken, and it is going very badly. Why do you think the entire world should double-down on this approach? http://energyforhumanity.org/wp-content/uploads/2017/11/European_climate_leadership_report_2017_WEB.pdf
Historical experience seems like it should carry some weight here. The only advanced economies that have decarbonised are those with lots of hydro and geothermal and/or nuclear power. Nuclear is a proven solution to decarbonised electricity. Solar and wind are not.
4. Your points only focus on electricity. But electricity and heat is only about 45% of emissions from fossil fuel combustion. Nuclear is much better suited to producing zero carbon fuels and district heating than solar and wind.
I think drawdown has lots of flaws as a prioritisation source.
(1) How they arrive at the ranking is unclear—the details on the models at the time I looked were very unclear.
(2) Technologies should be assessed as part of a whole system rather than individually. e.g. Having lots of energy storage makes sense when you have intermittent power sources like solar, but not much when you have controllable ones like gas with carbon capture. So, it doesn’t really make sense to assess the possible climate contribution of storage independently of everything else because its contribution depends on the whole system. Figuring out what energy system each country should have depends significantly on local context. e.g solar makes a lot of sense in Australia, but very little in England.
(3) I didn’t think that the potential contribution of various different energy technologies was well justified, and provided a false sense of certainty. E.g. nuclear could in principle supply the vast majority of global low carbon energy supply, but you have to think about all the potential unclear political barriers. Solar and wind could provide maybe up to 40% electricity, but you need to think about the massive land use implications of this and the consequent local opposition(4) They don’t talk about neglectedness, which is a crucial determinant of what difference donors can make on the margin. Wind and solar will probably be important going forward, but philanthropists are already ploughing hundreds of millions of dollars into advocating for them in Europe and the US. In contrast, things like CCS and nuclear get almost no money. Even less attention is paid to things like low carbon heavy duty transport.
(5) Some of the research seemed to be lacking in places. e.g. they put refrigerant management very high, but there are strong arguments suggesting that we should deprioritise short-lived climate pollutants. Similar thoughts apply to plant-based diets.
(6)The ranking is from the point of view of governments to a large extent. This isn’t a flaw but does make it less relevant for people donating.
Hi, thanks for this. I am personally very enthusiastic about CBT-i as it worked for me and has done for a couple of friends and the evidence seems to be reasonably good. One of the attractions of CBT-i for me was that you I don’t really think you need a trained therapist or an app, you can just follow instructions in a book about sleep hygiene and sleep restriction. I did use an online course, but I don’t think it was really necessary. The instructions are quite straightforward -
have a set waking up time. Spend the time in bed that you wish to spend sleeping (e.g. 7 hours) and then taper down time in bed by a half an hour each week until you reach the time you actually spend sleeping (e.g. 4 hours) and then ramp back up half an hour a week once the association between bed and stress is broken. If you can’t sleep for an hour get out of bed.
Sleep hygiene also seems quite straightforward—cut out caffeine, booze and nicotine, dim the lights, don’t look at your phone for tv two hours before bed, exercise. Just use bed for sex and sleep—don’t read in bed.
I think you could just get this by reading the second half of Colin Espie’s short book Overcoming Insomnia. Given this, what value do you think an app adds?
My favoured EA insomnia solution would be to publicise the Espie book far and wide and to get doctors to do what is required by their guidelines and recommend CBT-i, rather than just prescribing sleeping pills.
Thanks for these comments Alex. I agree that it would be best to look at how growth translates into subjective wellbeing, and I am planning to do this or to get someone else to do it soon. However, I’m not sure that this defeats our main claim which is that research on and advocacy for growth are likely to be better than GW top charities. There are a few arguments for this.
(1) GW estimates that deworming is the best way to improve economic outcomes for the extreme poor, in expectation. This seems to me very unlikely to be true since deworming explains almost none of the variance in economic outcomes across the world today, and research on and advocacy for growth looks a much better bet unless you endorse extreme scepticism about growth economics, which no EA has yet argued for. On the welfare metrics endorsed by GiveWell’s staff, deworming is roughly as good as their top charities. It is therefore very unlikely that GW’s top charities are better than research and advocacy for growth.
(2) The cost-effectiveness argument. Many of the huge growth episodes analysed by Lant occurred in countries that were extremely poor before those growth episodes. Looking to the past, it seems unreasonable to deny that funding research on and advocacy for growth is better than the best that one could do with a randomista intervention. The Chinese experience alone seems to me to clearly make this case. Looking to the future, our conjecture is that a 4 person year research effort will show that research and advocacy targeted at LMICs is better than the best GW charities. This takes account of the diminishing marginal utility of money. The case for this claim is unproven, but I think our argument provides strong support for it being probably true.
On the ‘risk-lovers would work on animals/long-termism’ point, I don’t think i agree. To me it seems that people work on these causes because of ethical assumptions about the weight of animals and future beings rather than because of attitudes to risk.
I agree that getting into the weeds is important for our predictive conjecture: the aim of our piece was precisely to motivate getting into these weeds. Moreover, someone needed to make these general arguments at some point as they had been around for many years without response.