Hacker News new | past | comments | ask | show | jobs | submit login
Let’s stop punishing IoT devices that embrace HTTPS (neosmart.net)
2 points by captaincrowbar on Aug 30, 2017 | hide | past | favorite | 1 comment



Just because it's an IP address in one of the private LAN ranges does not imply that the connection isn't being MitM'd, and it definitely doesn't imply that you're talking to who you think you're talking to. (How do I know I have the right 192.168.1.42? Am I on the "right" WiFi?)

The actual thing that TLS certificates set out to verify — that you're talking to who you think you're talking to — is simply not possible against a private LAN IP address.

(As an aside, I feel like it's entirely possible, technologically, to give IoT stuff unique DNS names, and thus, actual HTTPS. The problem is social, boiling down to the quality of the software we use. Consider: a machine, say an IoT device uses IPv6 (this is needed to get around NAT, since NAT won't work with the next bit). It can determine its hostname via a reverse DNS lookup on its IP address, and then, with ACME, get a certificate for that hostname. Now, it's not a pretty hostname, but nothing prevents ISPs from issuing better hostnames except a lack of imagination. E.g., I could be deathanatos.giant-isps-customers.com, my machines, machine1.deathanatos.giant-isps-customers.com, etc. Heck, the ISP could essentially delegate me that DNS space, or even allow me to set the reverse IP address record to point to my own DNS name that I pay for. I'm not 100% certain if it's entirely foolproof, but you could also envision a protocol for devices to allocate a subdomain under the customer's domain. But again, this will all never happen.)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: