4 ms·

> FBI say blast consistent with 552 tonnes of chemicals Seems unlikely that this estimate can have three digits of precision, which begs the question what the

> FBI say blast consistent with 552 tonnes of chemicals Seems unlikely that this estimate can have three digits of precision, which begs the question what the real margin of error is.

ThePadawan 5y ago
I actually don't find this to be unreasonably precise. I saw a video that showed how the blast was captured on around a dozen independent devices (smartphones, surveillance cameras) and how that data could be used to reconstruct a very good 3D capture of the blast and resulting smoke cloud. If the FBI spent X hours of simulation time to reproduce the 3D results captured on those videos, they might get it down to that precision (and who knows, maybe they just ran a lot of renders and 552 was the one that looked most right, while anything between 538 and 564 was also fine).
loufe 5y ago
Agreed, it does beg the question. Three digits of precision does not mean there isn't an error of +- 100 tonnes on the actual source calculation. I can imagine an engineer in the FBI did some modeling using damage radius, height of blast, etc. to get a rough estimate. News, being news, cuts off the error or the FBI, trying to look very competent, omits the error.
amelius 5y ago
Significant digits isn't always the best way to denote a number and its accuracy. If a simulation says 552 +/- 100, then how would you write down the fact that the expectation is centered around 552 and there is an error of 100, using a single number?
- contravariant 5y ago
  The best you can do with a single number would be something like 6·10^2 tons, you also use 6 hectotons or 0.6 gigagram though people might find those a bit weird. Or you could just denote its accuracy, which the original source likely does, but which news article inevitably fail to include (and subsequently fail to adjust the number of significant digits).
CamperBob2 5y ago
Usually this kind of thing seems to arise from a unit conversion by journalists working without an understanding of significant figures. The original report might have referred to "1000 cubic bloits" of nitrate, translating to 552 tonnes if you simply punch the numbers into a calculator.

> FBI say blast consistent with 552 tonnes of chemicals Seems unlikely that this estimate can have three digits of precision, which begs the question what the real margin of error is.

ThePadawan 5y ago

I actually don't find this to be unreasonably precise. I saw a video that showed how the blast was captured on around a dozen independent devices (smartphones, surveillance cameras) and how that data could be used to reconstruct a very good 3D capture of the blast and resulting smoke cloud. If the FBI spent X hours of simulation time to reproduce the 3D results captured on those videos, they might get it down to that precision (and who knows, maybe they just ran a lot of renders and 552 was the one that looked most right, while anything between 538 and 564 was also fine).

loufe 5y ago

Agreed, it does beg the question. Three digits of precision does not mean there isn't an error of +- 100 tonnes on the actual source calculation. I can imagine an engineer in the FBI did some modeling using damage radius, height of blast, etc. to get a rough estimate. News, being news, cuts off the error or the FBI, trying to look very competent, omits the error.

amelius 5y ago

Significant digits isn't always the best way to denote a number and its accuracy. If a simulation says 552 +/- 100, then how would you write down the fact that the expectation is centered around 552 and there is an error of 100, using a single number?

contravariant 5y ago
The best you can do with a single number would be something like 6·10^2 tons, you also use 6 hectotons or 0.6 gigagram though people might find those a bit weird. Or you could just denote its accuracy, which the original source likely does, but which news article inevitably fail to include (and subsequently fail to adjust the number of significant digits).

CamperBob2 5y ago

Usually this kind of thing seems to arise from a unit conversion by journalists working without an understanding of significant figures. The original report might have referred to "1000 cubic bloits" of nitrate, translating to 552 tonnes if you simply punch the numbers into a calculator.