I think they do only a hash per request. The 160*weight hashes were done per server (per feature set), to partition the hash space.
Per request you do a single hash and then a lower_bound on a sorted map to find the serving server (again, on the ring appropriate for the features required by the request, so likely a hash map lookup first)
You might have watched too much Chernobyl. It's not true that an irradiated person becomes itself radioactive.
Once a potential radiation contamination (radioactive dust and material) is cleaned from a person, independently of how much radiation they were exposed to, they don't become radioactive themselves.
You can easily look it up online to verify
The marginal cost of batteries grows more than linearly.
When batteries are covering 0% of the need, the marginally added battery cycles many times, so the cost is spread over many kwh produced.
When you add batteries to go from 99.98 to 99.99, the batteries cycle only for that 0.01, so the same cost to build them is spread over a much fewer kwh, making each kwh produced a lot more expensive.
Seasonal storage is madness: you charge once and discharge once per year. Pay 100$/kwh to install it, discharge 20 times (20 years, a 5% payback time, which is a bad investment), and you're paying those kwh a 5$/kwh premium on top of the cost of buying the discharged power.
If the battery is instead installed to shift the production from 12.00 to 18.00, it cycles 365 times a year, so in 20 years the premium is 0.01$/kwh.
So nucleare doesn't compete with the first 40% of penetration of renewables and the first 30% of battery, it competes with the last 10%, which is still needed to get to 0.
Per Ember Energy reports, a cost optimal new build grid is between 90% - 97% solar/wind/battery, and between 3% to 10% gas peaker depending on how much sun/wind your locale gets.
But you can't replace the gas peaker with 3% or 10% nuclear because in essence that gas peaker is supplying 100% of the power 3% to 10% of the time.
So you'd have to build nuclear plants that can supply 100% of the power. But once you have that you might as well use nuclear power 100% of the time because the rest is irrelevant. But that's about as far from cost-optimal as you can get.
As you mentioned, using batteries for seasonal storage is madness. So to get to 100% carbon free you have at least 2 other options:
1. use a different form of seasonal storage. China is experimenting with this. In 2026 they will double the world's pumped hydro storage capacity. But doubling isn't a lot -- the world doesn't have much pumped hydro. But it does mean they might start doing it at China scale in a few years.
2. Overbuild to avoid the need for seasonal storage. Solar works on cloudy days and in the winter. It just doesn't work very well. So you need a lot of it. Which is expensive, but still a heck of a lot cheaper than batteries for seasonal storage.
In reality, most places will probably say that 95% or 99% carbon reduction is good enough and keep their backup natural gas generators around for the occasional dankelflaute.
every single one of Ember's analyses are based on historical situations. That is certainly useful, and serves to demonstrate an important point - that wind and solar make economic sense.
There is not a single analysis out there that tackles the twin problems of meeting growing electricity demand with a power source that depends on an uncertain and changing climate.
You say "occasional" dunkelflaute, but we have no idea whether occasional is once a season or once a decade.
This is exactly where nuclear shines, because although it is expensive, it insulates a country against the vagaries of external forces. Whether or not that guarantee is worth the price tag is the question a lot of governments are grappling with and they mostly seem to agree that the value is indeed worth the cost.
> You say "occasional" dunkelflaute, but we have no idea whether occasional is once a season or once a decade.
Yes, these are things we know with a high degree of statistical confidence. Many locations have hundreds of years of hourly weather records.
We know that Europe has never had an hour where there wasn't significant sun or significant wind somewhere on the continent any hour in the past 30+ years. The same can be said for North America, and Asia.
What use is hundreds of hours of weather records when the climate is changing in ways we do not yet understand?
"somewhere" on the continent being windy is useless unless we have the transmission capacity to move that much power across thousands of kilometres. See also what just happened in australia last week - nearly 40% of the country stuck with very low wind speeds for nearly a week. What then?
I checked and it looks like ETS are increasing prices of gas produced electricity 20-30€/MWh.
So not little (although peaks are around 140€/MWh).
But these are taxes that can be used to reduce the reliance on gas (actually, it's mandatory to use them for transition projects).
It's unfair you're being down voted, you're right.
I used to think that we could get by with just solar wind and batteries, but then after collaborating with people on an ideal energy mix the numbers were obvious: there is a (small) fraction that cannot be covered.
Not with storage (the discharge cycles are so few that the cost is prohibitive. How can a battery pay for itself with 10-20 discharges a year? And this applies to any kind of battery that needs to be built, including hydro).
Likely there will need to be some baseload nuclear (which then increases average prices, since to make it economical you need to buy all the electricity it produces, and so it partially displaces renewables). The alternative is overbuilding solar+wind+battery something like 5/8 times the average need. Maybe if the prices drop enough that could be feasible..
The big win would be if there is some way to get predictable power at a lower cost than nuclear (e.g. tidal), which could be used to smooth the troughts, or alternatively a low capex but potentially high opex solution which is turned on only when needed (gas is an option, but not co2 free. And sizing the power needed is not super cheap, although now it's not a problem since we have enough gas capacity which is going to be displaced, so it won't be needed to be built)
Yeah but we are nowhere near the end of the scaling curve. For now, we can use the natgas plants during the unexpected outages while solving for green hydrogen / whatever backup plants. Like when a household has one EV and one gas car, they can always just take the gas car when they have range anxiety and don't know about chargers. NBD.
Net zero is barely enough to help with climate goals, given how late we are. It's not a huge goal, it is the absolute bare minimum to avoid >2 degrees of warming.
Achievable near-future net-nearly-zero in the near future is a lot better than waiting longer until we can achieve full net zero. Don't let the perfect be the enemy of the good.
The real issue is the cost of keeping gas peaker stations around that are mostly idle and fire up only a few days a year, but that's an economic issue, not an engineering one.
In the longer term, you could even run them off net-zero renewable syngas that you make the rest of the year using low-cost electrical power at peak solar generation times; you only need to store a relatively small amount of it, and old fossil fuel reservoirs are ideal for this.
> Likely there will need to be some baseload nuclear
Baseload nuclear is entirely feckless as a backup for a renewable grid. You either go with a long term storage technology (and then don't need nuclear), or you go to an entirely nuclear grid. Wind/solar and nuclear don't mix well.
Everyone who’s tried it suddenly realises that anything you put in the ocean is almost immediately covered in marine growth, or destroyed by the ocean itself.
And that wave / tidal energy is very diffuse, or that where it isn’t diffuse it’s also extremely destructive.
Capital costs starting to dominate doesn't mean that they increase.
If capital costs are 10 and operating costs are 90, but later operating costs become 1, the capital costs remain 10 and the price to recover the capital costs is the same.
Of course the capital costs need to be spread over less kwh, so the kwh will become more expensive, but what matters is the avg (and thus total) cost paid, and that would be lower.
E.g.
5B in capital cost for 25MW (amortized in 25 years), and 30 per MWh in gas cost.
If you produce 180TWh (20h daily of production) the average price cost will be (5'000'000'000/25 + 180'000'000*30)/180'000'000 = 31 per mwh.
If you produce 36TWh (5h a day) the average cost is 35, but the total cost is significantly lower.
At the limit, if the power plant produces 1MWh in the whole year, it will cost 200'000, but paying that MWh that high enabled the system to pay for all the rest of the year the electricity at less than 31.
You need to look at the total cost paid in the end by users, not at the marginal costs in some situations.
It's an auction, it's called marginal pricing. Every producer bids for x KWh at a certain price (for each time slot), and the cheapest y KWh to cover all demand are taken, and all are paid at the price of the most expensive KWh bought. There is plenty of economic research on auctions and why this system is optimal: this system incentives to bid at the actual marginal cost of producing electricity, and thus allows to discover the optimal price. EU has been discussing about changing it, but there hasn't been a better system proposed yet.
If you change to pay at the bid price, you'll have companies build teams of analysts to predict the market price and bit at that, they're not just going to accept being paid much less of what they could.
Instead of focusing on this, here a few more impactful things that would help: 1. Zonal pricing, so that there is an aligned incentive to build production where demand is (connection to the grid is a big limiting factor) 2. Stop providing contracts to renewable where curtailed production gets paid (curtailed energy is paid by consumers as taxes on bills normally) 3. Start allowing to build more renewable so that renewable are setting the marginal price 4. Push utilities to do PPA (power purchase agreements) with producers of RE so they can agree to a fixed price, and a smaller slice of electricity is bought at the auction
There are a few more, but these are the most important.
Regarding your edit: the gas plant is not subsidizing, the customers are paying. But of course at the moment building renewable is so lucrative thanks to this setup, that there is a big incentive to build them. Of course they need to plan for 20-30 years, and the risk of getting to big periods at 0 marginal pricing is real, so builders need to evaluate well the risk (and PPAs can help)
In a non-competitive world what you say would be true (you'd avoid filling in the remaining 1%), but there are a large number of power producers that a cartel is unlikely
Not only that, after X months of not paying the landlord can request an eviction, and from the moment the eviction is requested to the moment its executed the government pays the rent to the landlord (I believe 3 is reasonable for X, but it can be higher or lower).
I'm a renter and always have been a renter, but it has always been crazy to me that the cost of people without a house is dumped on the unlucky landlord.
If the decision is that this cost must be put on society, at least it should be spread among all citizens/landlords (e.g. with a tax on rent).
Not having the government paying while waiting for the eviction is the same as now: the government can just take ages.
The risk to landlords when the renters doesn't uphold the contract needs to be limited. This is also the reason why an insurance would be an inferior solution: there is still no incentive for the government to intervene fast.
We might find out that it's preferable to build social housing rather than paying rent to landlords, and the government starts actually doing something about it
Catholicism has been in general against contraception.
Of course it's still used, but probably at lower rates.
Non protected sex leads to pregnancy at a much higher rate than protected one.
I expect it to be similar for other similar religions.
Bodycount is also irrelevant, the total numer of intercourses is what matters. I'm not sure if having a stable partner since a young age leads to more total intercourses than occasional partners, but I'd be lead to thing so.
reply