From: Vincenzo lozzo < Sent: Monday, September 26, 2016 11:56 AM To: jeffrey E. Cc: Joi Ito; Danny Hillis; Reid Hoffman Subject: Re: Warning: this is=likely going to be a long essay, but I think it gives some perspective on t=e topic. The short version is: I was talking to Joi about this the other day -=1 wouldn't pay too much attention to this, Schneier has a long tradition of=dramatizing and misunderstanding things. That said, buying cloudflare (it's pr=vate) or akamai stock might be an idea because DDos attacks are not going a=ay soon and as a trend they will likely increase. Also to provide some perspec=ive, he links to this: https://www.verisign.com/assets/report-ddos-trend=-Q22016.pdf <https://www.verisign.com/assets/repo=t-ddos-trends-Q22016.pdf> If you look at the numbers on page 9 they are an o=der of magnitude smaller than the attack on a, rather unknown to the genera= public, cyber security journalist: https://krebsonsecurity.=om/2016/09/krebsonsecurity-hit-with-record- ddos/ <https://krebsonsecurity=com/2016/09/krebsonsecurity-hit-with-record-ddosh <=pan style="background-color: rgba(255, 255, 255, 0);">And we are talking a=erage not median.. CloudFlare was able to absorb most of the attack=on Krebs' website and they probably brought the website down because he wou=dn't be able to pay for the service at that rate anyway. So those numbers t=ere are nowhere close to "probing" the limits as he suggests.. =div> </=iv> The more i=teresting aspect is that DDos attacks boil down to two strategies: 1) "=mplication" attacks - which is a text book example of the tragedy of the co=mons 2) Force real traffic to happen. This often happens when you compromise a=device (that's what a botnet does). In that sense loT is particularly dange=ous (1) Is a good argument for Danny's idea of building a backup in=ernet. The problem there is that a lot of protocols (DNS, NTP etc etc) have=design flaws in that you can generate asymmetric amounts of t