It's funny how quickly that's gone from, "Oooo I can help you!" to "you guys are helpless."

I think the question is: If I have a 1000 packets and I lose 10 is the packet loss spread out of is it clustered? If it's clustered you'll have to retransmit anyway.

Interleaving the FEC can also avoid bursts. The tradeoff is the more you interleave, the higher the latency on reconstruction. They could do something like using round trip time to calculate the number of packets in the air and interleave based on that.

I was working on this type of stuff 15 years ago and now might be the time to do it. Bandwidth keeps increasing but latency stays the same, so it makes sense to waste some bandwidth to improve latency.

When you are trading at a P:E ratio of 50 you are expecting some phenomenal growth and at LinkedIn's size that growth is pretty hard to get. Scaling it back to 25 seems pretty damned reasonable honestly.

Eh, it's a different beast. Oliver only has to do one show a week and he has an hour. He doesn't want to just be John Stewart's show but with swearing. Doing this every night would be tiring for the audience. Doing it once a week saying, "hey here's this one fucked up thing in our society" is a digestible form.

To put it another way one is comedy news, the other is a comedy news magazine.

Of course, having a whole week to prepare and research the content really helps at quality as well.

What also helps is that John Oliver is on HBO - they have the budget, and they can swear or show pictures of genitals.

I have to give the CS rep some credit as he does seem to be trying to make things better. It's more response than a lot of manufacturers would have given.

> It doesn't fit on a simple chart with simple graphics.

Yeah it does just keep zooming in.

I've never understood why you wouldn't proxy the ads through the server providing the content. No more domain black holes.

Because the X,000 ad servers are far more technically competent at ad serving than the X,000,000 content servers.

I can't imagine the scope of a project that would move water 1500 miles over a mountain range. From the Missouri valley to western Kansas is a 3000 ft elevation change. You are literally trying to get water to flow uphill.

Making water flow uphill is more feasible than getting everyone affected by a substantial change in the Missouri River's flow to agree to such a diversion.

No kidding. If the whole damn Colorado River isn't enough to satisfy the West Coast, it will be a cold day in hell before the East shares their water.

They dug a channel through Panama by hand. We humans are pretty good at engineering.

The panama canal is 50 miles wide and has an elevation change of 85 feet. Construction took 10 years.

It would be vastly more efficient to have a pipe line which would still be ludicrous.

They should really try to recruit the survivors to do some of this work. They very likely have immunity now.

There are fairly basic steps to prevent HIV or Hep C. The worry has been if Ebola really gets loose and how quickly that could utterly fuck the world up.

I always thought the issue with Ebola was how quickly the host dies. If there were a period of about a week where no symptoms arose, then the virus would become significantly more deadly.

Until then, it doesn't seem probable that the host lives long enough to spread the virus.

Then again, my info is based off of what I researched in school around the mid 90s (fresh off of reading The Hot Zone).

"If there were a period of about a week where no symptoms arose, then the virus would become significantly more deadly."

This is precisely the problem with the new Ebola Zaire strain: the incubation period varies from ~3-21 days, during which the patient may show no symptoms yet can spread the virus. And since Ebola Zaire's mortality rate is 60-90% (early estimates were on the high end) this is an epidemiologist's nightmare.

I looked at the graphs and thought it might be time to go live in the woods for a bit.

Wow do you have a citation for that? Because that's utterly terrifying.