Posts

We as a society need to start wearing masks or this won't stop

Things will not start getting better until the infection slows down. That won't happen until either most people are infected or people start wearing protection in the limited social interaction: grocery shopping, essential working, etc. For various reasons, the CDC has recommended people to not wearing masks. However, things are changing and there are more evidence now that masks helps: Masks can reduce the amount of droplets you inhale and contact with your mouth and nose. It is never one virus that get you sick, it is the amount of virus that matters, reducing the amount of droplets will reduce the risk. There are more people infected every minutes and it grows exponentially, and most of infected people are asymptomatic. That means everyone, me and you, or anyone you come to contact remotely, however healthy one looks, can actually be infected and contagious. Masks protect others. Masks shows you care, and can be a sign of safety. If you have to take take-outs or foo

How many Starlink satellites will you see?

How many Starlink satellites will you see? According to the current plan , there will be ~12000 satellites on 3 orbital shells on height 340km, 550km and 1150 km. For estimation, it is good enough to assume the satellites will more or less evenly spread on each orbital shells. We need to calculate how many will be visible to one point on the ground for satellites on each shells. In the diagram left, the inner circle is the Earth. For all the satellites on the orbital shell h away from the ground, a person on the ground can see all the ones from the blue cap. So the ratio of visible to all satellites are the ratio of area of the blue cap to the area of the sphere. With  $h =\text{height of orbit}$ and $r = R_{earth} + h$ We have the ratio of visible satellites to all satellites on the same orbital shell: $=\frac{2\pi r h}{4\pi r^2}=\frac{h}{2 r} = \frac{1}{2}\frac{h}{R_{Earth} + h}$ With the number from the current plan: orbital shell Height # % visible #

5 secs fact check: does data center use more electricity for cooling than for computing?

No. Let's assume the energy used in datacenter are from 2 parts: computing(including all the power from the servers) and cooling. We know that for a cooling system(at least a decent one), the energy need to move the amount of heat: $W = qQ = T_{low} / (T_{high} - T_{low}) Q$ where q is the efficiency of the system, and almost always > 1 (only if the outside temp$T_{high}$ is extremely high already or we want to cool the system to super low temperature($T_{low}$); neither is true for data center) so no, for all the electricity we use for computing, it all turns into heat, and to move that heat out, we always use less than that energy.