About this interactive
Seven scenarios, seven cryptographic failures, one pairing each. The sorting question underneath the activity is not what each failure is called — it is who put the exposure there, because that decides who can fix it and when. Three of these are things an adversary actively does. A downgrade attack works on the part of a TLS handshake that has not been encrypted yet: a machine in the middle interferes with the negotiation so the two ends agree on an older protocol version or a weaker cipher suite than either would have picked on its own, and every byte from then on is protected at a standard neither party intended. The defense is not clever, it is arithmetic — the fewer old versions and suites a server still offers, the less room a downgrade has to work in, which is why a web administrator has to keep drawing that line and keep redrawing it as the standards move. A collision attack, also called a birthday attack after the probability puzzle it borrows from, targets the hashing algorithm instead of the connection: it exploits the likelihood that two different inputs produce the same hash output. Once that happens, the hash has stopped doing the one job it was there for, because a fingerprint that two documents share proves nothing about which document you actually have. A compromised key attack is the bluntest of the three. Someone has the key who should not — and if it is a private key, they can now impersonate the site it belongs to, which quietly cancels every guarantee that key was underwriting. Three more are not attacks at all. They are decisions, usually old ones. Weak cipher selection is a server still advertising an algorithm or suite that was defensible on the day it was configured and has since been deprecated or broken; nothing attacked it, the choice was simply never revisited. Insufficient key length is subtler, because the algorithm is fine and the implementation is fine — the key is just too short. Every bit you add multiplies the work of exhaustive search, so a key chosen too small hands back most of the cipher's strength for free and puts brute-force cryptanalysis back within reach. Self-signed certificates are the most human failure here: the certificate is vouched for by nobody but its own issuer, the browser objects, and users get trained to click through. The certificate is not really the problem. The habit is, because those same users will now click through a warning that an actual attacker triggered. The seventh belongs in neither pile, and it is the one worth thinking hardest about. Store-now-decrypt-later assumes nothing about today. An adversary intercepts encrypted traffic now and stores it, unread, on the expectation that the computing power to break it — quantum computing being the reason that expectation is no longer speculative — will arrive within the life of the secret. For a shopping session that is irrelevant. For classified material, medical records or anything else that must stay confidential for a decade, it means the encryption standard that satisfies today's threat model is the wrong standard to be measuring against. The takeaway is that "is our cryptography secure?" is an incomplete question. Secure against whom, and secure for how long.
About TechKnowSurge
TechKnowSurge builds IT and cybersecurity professionals through hands-on, concept-first training built around real understanding — not memorization. Free interactive tools, structured programs, and 25+ years of real-world experience, all in one place.
Explore free tools and programs →