And https://www.map.signalbox.io/, which tries to interpolate signal locations onto a geographic map (with the expected level of inaccuracy, though still not half-bad)
Surprised it allows FSD in such poor visibility conditions. Surely they need to do some work on a reported visibility confidence level where FSD disables if it is not confident in the amount of visibility it has.
Autosteer (lane following) will disengage if it can't see lane markings. I'm genuinely surprised FSD was active in these weather conditions (and as other posters have suggested, it may not be FSD at all).
Oh, the video capture device: because we had everything on analog CCTV, I had two analog TV tuner video capture cards in the server. Plain old 640x480 black and white analog video. When someone pressed the screen capture button on a Cocktail Console, I changed the channel on the video capture card to the appropriate channel, did a screen grab, and dumped the file in a folder on the server. People pressed it infrequently enough that two cards were fine to handle all the volume.
Every day I'd create a new mm-dd-yyyy folder for images to go to, and the Remote web site had a calendar on it. You could go to the site, click on the night you were there, see all the images captured by all people that night, and save your images if you felt like it.
Yep! First, here’s a video from the guy who developed all the hardware, Leo Fernekes. (He runs a great YouTube channel called Leo’s Bag of Tricks all about electronics and neat stuff you can do. Leo’s a genius.) Lots of details in here.
The Cocktail Consoles (as we called them) were all custom hardware. Everything was designed to be rock-solid both physically (bars are full of drunk people and liquids) and operationally (everything had to Just Work). Leo designed a core “motherboard” which was a PIC microcontroller (I forget the exact model) that did five main things: serial I/O for the buttons and joystick; serial I/O for the attached TV tuner; serial I/O for the attached pan-tilt video camera; audio from the telephone handset; and then multiplexing all of that serial I/O and sending it over serial to a central server (which I wrote — in Perl!) which then controlled all the Cocktail Consoles in the bar.
We used black and white cameras because they were both cheaper and also had much better sensitivity to low-light conditions (this has changed somewhat — but not entirely — in 20+ years) and black and white tube TVs because they were cheap. (This part was actually really dangerous — tube TVs hold enormous charges after they’ve been switched off, enough to kill someone, and we had the guts exposed on the insides of the Cocktail Consoles. Had to be very careful). We used public telephone handsets for the audio because of their durability, and video game buttons and joysticks so you could try very hard, and generally fail, to damage them.
The TV's, cameras, and telephone audio were all connected over an analog CCTV system. The camera was video source and the handset's microphone was the audio source for a given channel. The TV could be tuned to any channel, and was thus the video output device, and the handset's speaker was tied to the same channel. Thus, if you tuned to any camera, you would see and hear whatever was going on at that console, but not the other way around, so it was rather voyeuristic. If TV A was tuned to camera B, and TV B was tuned to camera A, that established a bi-directional link, which meant you could see and converse with the other person.
The serial data from all the microcontrollers were sent over serial-to-CAT5 converters, so the entire place was wired for Ethernet, but it was plain old serial over the wire. We then had these serial cards in a Dell server on the other end, which presented as roughly 100 serial ports on the server.
This was where I had to do a lot of learning. I was a good IP programmer, but I had to reach back into the depths of the kernel and learn all about TTYs and switch() and lots of other stuff that even in 2000 was sort of forgotten. It took me forever to find any good documentation on how to handle that many serial ports in a non-blocking way.
I kept asking Leo to just put a cheap Intel box in each machine and do it all over regular Ethernet, but he (rightly) kept insisting on this low-cost, rock-solid approach. Today the calculus would undoubtedly be different — you would do everything over IP — but back then Leo had a level of foresight I still admire.
Yep. 3rd and Bowery, before CBGB closed and the East Village went from the bohemian hipster world of RENT to the expensive place it is today. The Bowery had just barely changed from “don’t go there ever” to “oh, cool!”
For anyone looking to find out more about how computers work at a very basic level, I really recommend the book "Code" by Charles Petzold. Goes from first principles all the way up.
One thing that I find hard to understand is how the electricity prices in the UK have gone up so dramatically (blaming gas prices) when a large amount of the electricity is not generated from gas. Is the price being artificially inflated?
Electricity prices in the UK (and most other places) are set by the marginal unit, which is the most expensive unit that needs to be turned on to meet demand. All other generation for that time period gets paid the same price. The marginal unit in the UK is usually gas, hence the sensitivity to gas prices
What price do you think we should pay for the electricity?
Suppose we insist we'll pay less than the price you agree to sell for. Obviously that's not a sale, that's robbery. This problem arises even if we agree to pay everybody the average, because some suppliers didn't bid average, their bid was higher, but we still claimed their electricity, so we are stealing from them.
OK, suppose we decide we'll pay all accepted bids at their bid price regardless of the marginal unit cost. If we do this the supplier is incentivised to guess the bid we will accept, so as to collect the difference between their actual price and the price we're willing to pay. If they're very good at this, we pay exactly the same as now, but, regardless of whether they're good at it the grid is significantly destabilized by the increased uncertainty due to lack of efficient price signals.
I don’t know much about it; that’s why I asked the question.
So the short answer is I have no idea, but I doubt we currently live in the best of all possible worlds.
It seems an odd pricing scheme which isn’t working well as gas becomes less desirable and renewables are a bigger part of the mix.
What you describe as paying at the bid price sounds like a free market to me - if you’re going to have a market mechanism perhaps it should use the market to efficiently discover prices, not impose a very odd pricing structure - if imposing nationwide prices, why have a market at all?
There are many possible alternatives:
Allow prices to float in a completely free market, and force suppliers to hedge (which they do anyway to some extent, but they could be required to).
Nationalise the national grid and energy production
Build interconnects and hydro storage to reduce reliance on gas plants and reduce the problem.
Set prices in regional markets to bring production closer to the areas with peak consumption.
Set prices by energy type (separate markets per type like fusion solar etc taking into account externalities).
Set prices by energy use (baseload, fluctuating, peak)
Defining pricing mechanisms is not robbery it is why we have a regulator!
The reasoning is that it incentives electricity producers to offer max amount of electricity at low prices without speculating how to maximize profit (as their sell offer will practically speaking have zero effect on the spot price). Nuclear plants, wind power, solar can just offer to sell at everything at around 0c/kWh.
It's claimed that another type of market would cause companies to speculate with their sell offers and thus generate less electricity. It would be interesting to see how this kind of market would work in reality, though.
There are many possible ways to decide (by region, by use, by type of energy production). At present we all pay exorbitant prices because the system is badly designed and badly planned.
My understanding is most wind was bought at a guarenteed price by the government at the time of construction, so a wind farm producing 1MWh gets paid say £40 regardless of the cost of electricity on the grid - even if marginal cost was £20/MWh
As users are then paying £90/MWh for gas, does the excess £50 go to the government or to the wind far owner?
The government. The mechanism is called Contracts for Difference, and as the name implies they work by ensuring the difference between an agreed strike price and the actual price - in either direction is paid
However, notice two further considerations:
1. Such contracts eventually expire. Exactly when varies. But the wind farm is still there, just now the energy price all goes to the operator.
2. Older government subsidies were not CfD. Ten years ago if you built a wind farm you got a direct subsidy. The CfD schemes come into existence from about 2014. They're one of a small number of good ideas the Tories had. They're in line with Tory ideology, but they also actually make sense in the world that actually exists.
The price of electricity in the UK is linked to the most costly supply in the entire mix. So if gas is the most expensive then we pay all other power-producing suppliers, regardless of means of generating, the wholesale price we'd pay for gas. It's a strange system.
The marginal price of electricity (i.e. the price of the most expensive source) is what drives the retail cost, because it's a liquid commodity that can't (to a first approximation) be stored. Imagine 90% of your electricity comes from wind and the last 10% has from gas because there is nothing else - the price of electricity is going to equal the price of gas because the wind providers can raise prices until they're just under the price of gas, since there are no other options. The most expensive form of electricity sets the price until it isn't needed anymore and is booted off the grid entirely, but once you cover that last 10% with wind then the price falls dramatically.
In theory this is what we want: the windfall profits on cheap power during periods of expensive energy are supposed to attract the market to build more of these plants and chase those profits, thereby accelerating the green transition. But it's possible what we saw last year was too much, and that the damage to the economy (nothing strangles economic growth like expensive energy) does more harm than this incentive does good. People are talking about renegotiating power agreements in Europe to pay fixed prices for renewables so this wouldn't happen again, but I haven't heard how likely this is to succeed.
The way it works in the EU—not sure if the UK participates in this market still?— is that every energy source is priced according to the most expensive energy source until demand is covered by supply.
So, for example, on particularly windy days here in Denmark, we pay almost nothing as our entire demand will be covered by wind energy. On other days we might pay a lot since we need to import energy produced from gas or other expensive sources.
There's a couple reasons. A small amount of electricity may be generated from gas, but that small amount is often crucial and in a lot of demand. Furthermore, peaker plants will produce less and less electricity as more renewables reduced their duty cycle. But the overhead cost of maintenance remains: reducing a gas plant from running 10 hours a day to 2 hours a day does not result in a 5x reduction of cost.
The whole system as detailed in the article seems pretty artificial and not great. For some reason prices appear to be set at the national level, ignoring the fact that Scotland has an excess of wind energy. If consumers could see that the price difference on their end, I guess there’d be more incentive to upgrade the infrastructure and get it down to England more efficiently.
If gas is setting the price as the most expensive form of energy, then it acts as an incentive to build cheaper forms of energy because your margins are that much higher.
Alas, England doesn't allow on-shore wind power, and there's not sufficient capacity (in terms of HVDC lines) to transfer enough power from Scotland down to England to move all of the excess energy.
That seems like a good way to spur early development. I wonder, though — if consumers could actually see the cost benefit of the wind power, might gas have been just priced out of the system by now? (Or relegated to some backup status). (Supposing the transmission infrastructure were upgraded to allow for the higher flow, or England changed its laws to be more in line with economic realities).
No, first understand the concept of market clearing auctions[1], then understand that the there is a dispatch stack (that looks like something like this [2]), and that gas plants are the marginal producer required to balance the market, as they are not baseload (nukes) and not intermittent (renewables) but are dispatchable (ramp up/down capacity as need to balance the market).
You can't base your electricity prices solely or even mostly on an unpredictable source of generation, which is nearly absent one day and generates more power than is needed on another day. Efficient storage is a long-term fix for this, but it ain't here. Natural gas is the most flexible source of on-demand power, so it (disproportionately even to its share of 1/3rd of all generated power) affects consumer electricity rates.
A couple possibilities. It could be competition from other countries with a different mix (e.g., the price in the UK has to compete with other countries or it'd be exported to the degree possible)
Power is also sold (ok, not sure about the UK) at the market price. So the most expensive generation needed to meet demand determines the market price. So even if gas is a small portion of the generation it could still determine the price.
If you haven't looked into it yet - Google Apps Script [0] is an incredibly powerful system when combined with Google Sheets. You can integrate lots of business processes and external systems (rest,soap,databases etc) together and integrate with classic spreadsheet functionality, we use it a lot in our business!
Yeah, I had a similar idea to Rows, but decided it'd be better to add what's missing to Sheets vs. build an entirely separate app. Wax (https://www.wax.run/) helps you build internal tools on top of Google Sheets. It's built using Apps Script and handles a ton of the boilerplate you'd want for any internal tool. Scheduling and integrations are hard to do in Apps Script, so we handle that and considering a lot of internal tools leverage SQL and Python we make using those them from Sheets simple.
App Scripts is great and together with Sheets it almost feel like a tiny functions + db serverless engine.
Having used it significantly, I found two main qualms that prevented me from investing more in the platform:
1. It is just JS syntax but the runtime and the “standard library” are completely different than node.js. For example there’s a weird UrlFetch class instead of xhr/fetch. It’s a never ending learning curve so much it feels like another language. Any async support is also non existent, that means no setTimeout, no Promise, no async/await (unless they added it recently with the move to ES6)
2. Due to 1, the library ecosystem is very limited. A library in GAS is just another GAS script that you import by referencing its ID (the long one in the URL).
Scripts can be public so you can import other people’s code. However no npm, no lodash and friends.
You can dev locally using clasp, so you can use git/npm/.., and package your code with webpack or something so it runs on GAS. However this only works for npm libs that do not depend on the usual JS environment and APIs. Lodash will work, anything network related won’t for example.
All in all Google App Script is more of a hassle than it’s worth so I only use it in specific cases, but I wish for a similarly accessible sheet+code environment that can also serve HTML, just with a better JS runtime.
Oh and don’t get me started on the development/deployment lifecycle.
I've been working with this for a while and I find it frustrating slow and unreliable. Especially working with an HTTP listener. Does anyone have better experience/advice?
Yes, don’t use GAS if you have the possibility to avoid it.
Otherwise, do the heavy lifting somewhere else and use GAS as a very basic wrapper for your sheet.
One pattern that worked well for us in the past is to use Firebase functions (or Google Cloud Run) and leave the scripts to act as a dumb API to interact with the sheet.
Is there a solution to version control/deterministically see/deploy active scripts? How do you manage all the scripts? I've used it to compile/summarize spreadsheets and serve as a simple endpoint (testing webhooks), but not much more.
