February 13, 2026
Big‑O vs Big‑Oh‑No
Faster Than Dijkstra?
Internet nerds brawl over “faster than Dijkstra” — genius or just big‑O bragging
TLDR: A new algorithm claims to beat the classic Dijkstra approach for shortest paths by avoiding sorting, impressing theorists. The comments erupt with skepticism, sorting nerd fights, and jokes—most want real-world benchmarks and ask if it matters for networks that aren’t massive, not just prettier math.
A new mathy routing trick says it can beat the classic Dijkstra method for finding the shortest path across networks by “breaking the sorting barrier.” That’s spicy, since Dijkstra’s 1959 algorithm is everywhere, including OSPF (Open Shortest Path First) used by internet routers. The paper passed top‑tier peer review, but the community asks: does theory matter if your network is only hundreds of routers, not millions?
The loudest camp: pure skepticism. User qsort shrugs, saying it’s “better big‑O, worse in practice,” and folks pile on with examples where prettier math loses to gritty real‑world code. Then the thread combusts into a sorting showdown: shiandow flexes that some data can be sorted in linear time, while vprcic relives a “lively debate” with a boss over whether counting sort proves sorting is O(n). It’s part computer‑science lecture, part family drama.
Meanwhile, meta chaos: alias_neo calls Deja Vu on the post (“they do textbooks?”), and jason_s derails into font beef—Epilogue drives him wild, but his employer blocked extensions, so no fix. Memes fly: “breaking the sound barrier” jokes, and lots of “show me benchmarks, not proofs.” The vibe is clear: cool theory, but wake us when routers run faster.
Key Points
- •A new shortest‑path algorithm claims better asymptotic performance than Dijkstra by avoiding sorting (“breaking the sorting barrier”).
- •Dijkstra’s algorithm, with complexity O(n log n + m), remains the basis for OSPF and IS‑IS implementations per specification guidance.
- •The new approach claims O(m log^(2/3) n), which is asymptotically smaller but may not be faster for typical network sizes due to constants.
- •The article emphasizes that real‑world usefulness depends on practical scales; smaller networks may favor established methods despite worse asymptotics.
- •Anecdotes (Bellcore vs. FORE Systems) and the supertanker analogy illustrate that scalability benefits must be weighed against practical deployment needs.