Designing Better Artefacts
@pmholling.bsky.social
Senior Lecturer in Aerospace Engineering, expert on uncertain decision making- Retweets are things I find interesting, but do not, always reflect my opinions.
I’m very sorry
January 27, 2026 at 5:00 AM
I’m very sorry
Subsidy spend in this case would be the net flows, via the Low Carbon Contracts Company from rate payers, through suppliers, to CfD covered generators. The CfD levy is applied directly to each MWh that energy suppliers procure. So all in you payments to your supplier.
January 20, 2026 at 3:06 PM
Subsidy spend in this case would be the net flows, via the Low Carbon Contracts Company from rate payers, through suppliers, to CfD covered generators. The CfD levy is applied directly to each MWh that energy suppliers procure. So all in you payments to your supplier.
Might be missremembering but I seem to recall that was the working estimates for how long a project started now would take to plan build and commission.
January 16, 2026 at 4:40 PM
Might be missremembering but I seem to recall that was the working estimates for how long a project started now would take to plan build and commission.
The work up for DESNZ on gas had delivery timeframes of after 2032 and LCOE at ~£105/MWh for 90% load factor, and my understanding is that the CAPEX values that was based off of were estimated by some stakeholders to be <1/2 what they believed them to be.
January 15, 2026 at 6:09 PM
The work up for DESNZ on gas had delivery timeframes of after 2032 and LCOE at ~£105/MWh for 90% load factor, and my understanding is that the CAPEX values that was based off of were estimated by some stakeholders to be <1/2 what they believed them to be.
Note: if we are going to judge decisions based on current knowledge then our entire existence is ‘bad’. On the metric the ‘decision’ to develop agriculture, in the aggregate is the worst decision humanity ever made. So bad, that it becomes virtually impossible to make a worse one.
January 15, 2026 at 7:17 AM
Note: if we are going to judge decisions based on current knowledge then our entire existence is ‘bad’. On the metric the ‘decision’ to develop agriculture, in the aggregate is the worst decision humanity ever made. So bad, that it becomes virtually impossible to make a worse one.
If the answer is yes then it is a bad choice. If the answer is no then it is as good a choice as any other.
January 15, 2026 at 7:17 AM
If the answer is yes then it is a bad choice. If the answer is no then it is as good a choice as any other.
It is entire reasonable to expect that we will pay more overall because we chose to ensure capacity is there in 2030/31. That is risk aversion. The fact that we pay likely more is a good thing. The question is are we paying more for no gain in certainty (based on our current beliefs)?
January 15, 2026 at 7:17 AM
It is entire reasonable to expect that we will pay more overall because we chose to ensure capacity is there in 2030/31. That is risk aversion. The fact that we pay likely more is a good thing. The question is are we paying more for no gain in certainty (based on our current beliefs)?
So you can basically count on reactors being available unless on scheduled mtc when called on. Solar, wind, etc may not be. The big issue is that nuclear is not complimentary the rest of the time, unlike stored hydro (both pumped and natural) or other tech. Even the silly H2 is complementary
December 30, 2025 at 8:38 AM
So you can basically count on reactors being available unless on scheduled mtc when called on. Solar, wind, etc may not be. The big issue is that nuclear is not complimentary the rest of the time, unlike stored hydro (both pumped and natural) or other tech. Even the silly H2 is complementary
Nuclear’s ‘advantage’ over many other low carbon sources, and more specifically smaller reactor designs is in availability. That is whether or not you can generate electricity when you need to. The outages of nuclear plants tends to be relatively uncorrelated, at least in the mid-life period
December 30, 2025 at 8:38 AM
Nuclear’s ‘advantage’ over many other low carbon sources, and more specifically smaller reactor designs is in availability. That is whether or not you can generate electricity when you need to. The outages of nuclear plants tends to be relatively uncorrelated, at least in the mid-life period
That said we found GPs were a nice go to when data was expensive and prior assumptions of behaviour were reasonably strong.
December 21, 2025 at 8:47 PM
That said we found GPs were a nice go to when data was expensive and prior assumptions of behaviour were reasonably strong.
Very true the non-parametric is always something people have trouble wrapping their brains around. “What do you mean you need to optimise the hyper-parameters, it’s non-parametric”.
December 21, 2025 at 8:47 PM
Very true the non-parametric is always something people have trouble wrapping their brains around. “What do you mean you need to optimise the hyper-parameters, it’s non-parametric”.
People don’t tend to thing about how much information is in the assumptions. Years ago when working on an insane surrogate modelling task using GPs we play around with the kernels for simple functions and in some cases it required >10 times the data to replicate.
December 21, 2025 at 9:39 AM
People don’t tend to thing about how much information is in the assumptions. Years ago when working on an insane surrogate modelling task using GPs we play around with the kernels for simple functions and in some cases it required >10 times the data to replicate.
The fact that CCGTs already know to bid just below the marginal cost of the next marginal supply should be obvious.
December 12, 2025 at 8:08 AM
The fact that CCGTs already know to bid just below the marginal cost of the next marginal supply should be obvious.
One other thing, ECO and WHD are the only policies spread over both has and electricity (though I would argue not done intelligently) removing ECO discourages electrifying, which has significant efficiency gains.
November 6, 2025 at 7:56 AM
One other thing, ECO and WHD are the only policies spread over both has and electricity (though I would argue not done intelligently) removing ECO discourages electrifying, which has significant efficiency gains.
The only sensible thing in that article is the idea of moving from RPI to a CPI based deflator. The continued existence of RPI is an embarrassment for ONS and HMT.
November 6, 2025 at 7:52 AM
The only sensible thing in that article is the idea of moving from RPI to a CPI based deflator. The continued existence of RPI is an embarrassment for ONS and HMT.
If I understand it correctly there is a floor in the revenue you receive, with subsidy if the market goes below that, and a cap in your revenue where you pay back above that. Kind of like a CfD but with a spread in between.
November 6, 2025 at 7:39 AM
If I understand it correctly there is a floor in the revenue you receive, with subsidy if the market goes below that, and a cap in your revenue where you pay back above that. Kind of like a CfD but with a spread in between.
It’s been a decent while since SVT was a reasonable tariff. It should be at its peak for reasonableness compared to fixed (gas wise) for Q1 and most unreasonable for Q2.
October 29, 2025 at 12:07 PM
It’s been a decent while since SVT was a reasonable tariff. It should be at its peak for reasonableness compared to fixed (gas wise) for Q1 and most unreasonable for Q2.
Those are not mutually exclusive. Heck, if he’s the only one with the wherewithal to ask follow up questions he ends up being, at least within the group, unique.
October 28, 2025 at 6:05 AM
Those are not mutually exclusive. Heck, if he’s the only one with the wherewithal to ask follow up questions he ends up being, at least within the group, unique.
BSUoS = balancing services charges
DUoS = Distribution network charges
TNUoS = transmission network charges.
DUoS = Distribution network charges
TNUoS = transmission network charges.
October 27, 2025 at 3:53 PM
BSUoS = balancing services charges
DUoS = Distribution network charges
TNUoS = transmission network charges.
DUoS = Distribution network charges
TNUoS = transmission network charges.
Much of rise of late is driven by RPI in many of the social and environmental policies, plus network costs. There was a big change back in 2022? when the DUoS charging model changed. Right now the BSUoS charges are quite high as the expectation of costs for this time a year ago was much higher.
October 27, 2025 at 1:59 PM
Much of rise of late is driven by RPI in many of the social and environmental policies, plus network costs. There was a big change back in 2022? when the DUoS charging model changed. Right now the BSUoS charges are quite high as the expectation of costs for this time a year ago was much higher.
I think the CfD interim levy is set to go down slightly in Jan, but could be misremembering, obviously the RAB interim levy is to come in. BSUoS, DUoS, and TNUoS are fixed until April. Then we are set to get a big jump in TNUoS chargers from ~13p/day for domestic consumers to ~25p/day.
October 27, 2025 at 1:24 PM
I think the CfD interim levy is set to go down slightly in Jan, but could be misremembering, obviously the RAB interim levy is to come in. BSUoS, DUoS, and TNUoS are fixed until April. Then we are set to get a big jump in TNUoS chargers from ~13p/day for domestic consumers to ~25p/day.