Hacker News Reader

Comment by shmerl | Hacker News Reader

That's 54.4°C for the context.

The title should be edited by a moderator to at least add the unit, this feels like a clickbait otherwise.

I saw 130 and assumed from the context it wasn't in the order of 40 degrees hotter than the highest temperature ever claimed to have been recorded

what about wet-bulb temperature?

If the relative humidity is low enough, it might be 25°C or less.[0]

0) https://en.wikipedia.org/wiki/Psychrometrics#/media/File:Psy...

It's an EXTREMELY dry place (around 5%), so the web-bulb temperature is not especially high, an online calculator estimates numbers in the low 30s Celsius.

A wet bulb temperature of 35 Centigrade is fatal within hours even with shade and fan cooling.

"Low 30s" wet bulb is a very big deal.

Sourcing: https://en.wikipedia.org/wiki/Wet-bulb_temperature#Wet-bulb_...

Well, it's not called "Death Valley" because of the landscape.

Based on reanalysis data, it looks like wetbulb temperature never goes above 25C around Death Valley. Average temperature seems to have gone up by about 2 degrees Celsius since 1980 though, from ~19.5 degC to ~21.5 degC.

Ref: ERA5 data & https://climate-explorer.oikolab.com

Really, we should be measuring max temp by wet bulb. There are places that are literally fatal to sit outside in, but Death Valley ain't one of them.

I will say it like this. I would spend 24 hours in death valley than 2 in Singapore...

Thanks!

What about in kelvin!

Celsius to Kelvin is easy: just add 273.

327.55 K

Oh no, not this again!

https://news.ycombinator.com/item?id=24118715

If the diameter of a coin is 19 mm, the circumference is not 59.6902604182 mm

I doubt your coin's diameter is exactly 19mm. By omitting digits after the decimal you are misleading people into believing that your number is more accurate than it really is.

If you round 327.55 K to 328K and convert that back to Celsius you end up with 54.85°C instead of the original 54.4°C. Increasing inaccuracy by half a degree for the sake of inaccuracy is bad science.

> By omitting digits after the decimal you are misleading people into believing that your number is more accurate than it really is.

Digits omitted at the right mean we don't know them, not that they are zero.

> and convert that back to Celsius you end up with 54.85°C instead of the original 54.4°C.

Yes, inaccuracy accumulates when you convert back and forth. But it's common practice in physics not to increase the number of significant digits to not give the impression a measurement is more precise than it actually is.

In this specific case: https://en.wikipedia.org/wiki/Significance_arithmetic#Multip...

And when precision really matters, then one should explicit the range (eg. +/- after the absolute unit)

It wasn't 54.4°C though! The original value was only 130 °F which is probably between about 54.2 and 54.7 °C but somebody made it look more accurate during that first conversion which fooled you with the very thing you're saying is OK.

You're bound to sometimes lose information doing round trip conversions no matter how many digits you have so don't do that if you care about keeping as much precision as possible.

The Guardian reported it as 129.9 °F because they converted twice!

> 328K and convert that back to Celsius you end up with 54.85°C

No, that's wrong. If you convert 328K to Celsius, you get 55C. You can't add arbitrary precision to a number. The correct calculation for a whole degree is:

54C => (54 + 273)K => 327K

If you want a single decimal place of precision:

54.4C => (54.4 + 273.2)K => 327.6K

You can then subtract 273.2 to convert back, and you get the 54.4C you started with.

I find it a bit amusing that we treat as gospel the use of in-band signaling to try to cram a half-assed error propagation into the value we're trying to convey. It's not a terrible system, but there's nothing inherently better about throwing away a modest amount of information about both a value and its error than throwing away a negligible amount of information about a value and a majority of what we know about its error.

Yea it's not nature's way, just a convention. But it's an intuitive convention because it's easy to assume that digits which are shown are correct and digits which aren't shown are unknown. The alternative - throwing away the majority of information about the error would make lots of numbers pretty useless. Is 130 F really 100 F with too many digits?

and in British Thermal Units?

BTUs are energy, not temperature.

Reply on news.ycombinator.com