Full Net Census Takes a Hint From xkcd

netbuzz writes "The University of Southern California Information Sciences Institute says it's the first full census of the 'visible Internet' since David Smallberg canvassed a piddling 315 allocated addresses in 1982. They're talking about 3 billion pings directed toward 2.8 million addresses over the course of 62 days. Oh, and they credit the comic strip xkcd for sparking the idea of presenting the data using a Hilbert curve." The main page for the census project has links to versions of the census at various scales.

Re:See also xkcd comic "Online Communities"

[Kandenshi]

For the lazy it's just to the "southwest" of the IRC isles, southeast of wikipedia. Easier to see in the blown up version of the strip here [xkcd.com]

Re:See also xkcd comic "Online Communities"

[SighKoPath]

It's a bit out of date, facebook expanded a lot. Myspace shrank and wheres slashdot.

On the Viral Straits and Bay of Trolls, between Reddit and Soviet Russia.

Re:See also xkcd comic "Online Communities"

[godscent]

It's a bit out of date, facebook expanded a lot. Myspace shrank and wheres slashdot.

In the Ocean of Subculture, south of Digg, bordering Reddit and Soviet Russia, is "/."

nmap

[blhack]

PSH..

nmap -sP *.*.*.* > ips.txt

Re:See also xkcd comic "Online Communities"

[gerbalblaste]

nope thats not slashdot. slashdot is labled /. and is bordered by reddit and soviet russia on the viral straights.

The isle of slash is something very different from slashdot, mostly involving harry potter...

Re:See also xkcd comic "Online Communities"

[Kandenshi]

You can find out (more than you wanted) what the deal with slash here [wikipedia.org]

Hint: It has something to do with codes like Kirk/Spock or Harry/Draco

Re:Really useful for the colorblind

[Anonymous Coward]

Did you notice the number of shades they used? Putting it in grayscale would give shit for detail.

Re:God, STFU

[adamziegler]

Even though you too are red green colorblind... it does not mean that you are seeing it the same way as the parent. http://en.wikipedia.org/wiki/Color_blindness#Congenital_color_vision_deficiencies [wikipedia.org] Besides... its fairly simple to design a website or chart that makes things easy for even those who are color blind. I too am color blind... Honestly, I can't tell if the negatives and positives are mixed together when I look at it. (Not that I am fluent in reading the chart anyway!)

colorblindness IS fairly comon

[Matthew Bafford]

The main similarity being, of course, that both color blindedness and .. blindedness .. are rare enough that the designers of the image hadn't even thought of it.

Stats vary (and you can look them up easily enough), but the general idea is that 1/12 males are color-blind to some degree. That means most groups are fairly likely to have at least one color-blind person in them. Now the severity of color-blindness as well as the affect that has varies significantly from one color-blind person the next.

I, for example, am color-blind, but didn't find the chart to be horribly difficult to use. Different colors might have made things easier, but it doesn't bother me in this case. That doesn't mean it shouldn't be considered when designing. Like I said, most color problems are due to plain ignorance as to how common the problem really is. I don't blame people for not considering it, as long as they really didn't realize.

Re:Why has nobody commented on the Hilbert Curve?

[MenTaLguY]

News bulletin: two points that are close to each other on a line are close to each other when the line is curved.

The Hilbert curve preserves that locality better than other sorts of space-filling curves, however.

Re:See also xkcd comic "Online Communities"

[Eq 7-2521]

There's even an xkcd about slash: http://xkcd.com/305/ [xkcd.com]

Play nice, let your firewall answer all pings.

[Medievalist]

You won't find any of my servers/boundaries responding to a ping on any address at any port for any reason. Send a TCP packet, and all of them will look at it, stroke their chins for a few microseconds, and decide whether to forward them or simply move on.

Are you sure that's all they are stroking? Just kidding. It seems a bit unnecessary to shut down your site's ability to help others test connectivity to you. You really aren't doing anything but crippling harmless diagnostics; it's very easy to make your network safe to ping.

A ping test is perhaps one of the silliest, as you cite by a more accurate observation of key SOA servers over a period of time.

But, you see, there's no single trustworthy authority that has root access to all the nameservers. Think about how DNS works, and how the hints file interacts with local and intermediate caches, and you will see that your idea is not really any more workable than a ping test. It's too impossible to co-ordinate. I cache at three levels for good solid reasons not having anything to do with "fear of a bad ping". On the other hand I assume pings are friendly and only monitor them for performance and bandwidth reasons, and I have not yet been hacked despite many years of pen tests by outside agencies we've hired.

That said, I like Novell.com's bravery, as they always respond to a ping. It's how I know that my DNS infrastructure is working. It's a randomly successful find (I have no affiliation with them), rather it always works, when it works.

Aha! You admit that your fears are impacting your ability to serve the community - in a way that you admit is valuable! This admission is the first step to great power! OK, just kidding again.

Configure your firewalls to respond to all inward-bound pings for your entire address space. This will not consume any significant resources, and will not inform any skeery crackers of anything (in fact it's a better way to fool them than blocking ping, since they will not need to resort to stealthier scans that require more resources to detect or block). Log who pings you to the router console and leave a dumb terminal running on it, or pump it into a secure internal web page. Treat ping flooding like any other kind of packet flooding - you can't really make it impossible to DDOS you simply by blocking specific ICMP types anyway. Don't forget to implement packet source ingress and egress filtering, obviously.

Google, yahoo, and Novell all respond to ping. It's a service they kindly provide to the rest of us, a service we should all provide to make the Internet's tubes easier to see through. You aren't going to get hurt by a ping unless you have no idea how to set up a network... in which case dropping ping packets won't save you.

Don't make researchers have to develop new ways to punch through firewalls, let's all just use good ol' friendly, simple, and useful pings.

93% non replies!

[Anonymous Coward]

How useful is it to know that 93% of all IP addresses pinged are non-replies?