War of the workstations: How the lowest bidders shaped today's tech landscape

Original Article ↗ Hacker News Discussion ↗

Cheap, “good enough” IBM PCs, DOS, and Unix systems are argued to have won out over more capable but expensive platforms like Lisp machines, Smalltalk environments, and proprietary workstations, shaping today’s largely homogeneous computing landscape. Commenters explore how openness, standard buses like ISA and EISA, and low-cost commercial off‑the‑shelf parts encouraged massive software reuse and ecosystem growth, while richer, more malleable systems struggled with portability, tooling, and market fit. The exchange raises broader questions about technological inertia, “worse is better” design trade‑offs, and whether there is still room to revive more powerful, user-malleable computing models in an industry now dominated by commodity hardware, Unix-like OSes, and the web.

IBM PC, Cost, and OS Competition

  • Debate over whether the IBM PC “won because it was cheap.”

    • Some note it was expensive vs CP/M/Z80 and 8‑bit home machines, but relatively cheap as a business desktop, especially when bundled with low‑priced MS‑DOS.
    • MS‑DOS undercut UCSD p-System and CP/M‑86, and its combination with open, cloneable PC hardware was decisive.

  • Digital Research’s Concurrent DOS (a multitasking, DOS‑compatible CP/M‑86 successor) allegedly depended on a 286 feature that Intel removed before release, crippling its chance to compete.

  • IBM’s open documentation of the ISA bus is seen as crucial to third‑party ecosystems and PC dominance, driven partly by antitrust constraints.

Evolution, “Lowest Bidder,” and Inertia

  • Many agree that simpler, cheaper systems repeatedly displaced richer, more sophisticated ones, but argue this is more “good enough for now” than purely “lowest bidder.”

  • Some see a convergence to lowest common denominator tools (e.g., C) that scale socially: easy enough for many “average” programmers, even if theoretically inferior.

  • Others counter that not all domains follow this: mainframes, bespoke industrial systems, and some research OSes pursued “better” rather than “cheaper” and survived in niches.

  • Inertia and 30+ years of accumulated software are viewed as the biggest barrier to “starting over” with better paradigms.

Memory Limits and Architecture Transitions

  • Analogy drawn between 64KB limits on 8‑bit CPUs vs 4GB limits on 32‑bit x86.
    • Critics say they’re not comparable: 4GB rarely blocks normal workloads, and PAE mitigated system‑wide limits.
    • Proponents argue that, in both eras, early adopters hit limits first, pressuring a transition before the mass market felt it.

Lisp, Smalltalk, and Dynamic Systems

  • Strong disagreement over “Lisp fetishism.” Some criticize over‑romanticizing Lisp; others defend its historical importance but prefer successors like Dylan or Smalltalk‑like systems.

  • Several describe Lisp and Smalltalk environments (especially Lisp machines) as fully live, introspective, self‑documenting systems that far exceed today’s browser‑based setups.

  • Others found Lisp machines complex and hard to learn without guidance, highlighting a steep learning curve and “weirdness” for most users.

  • There’s speculation that some people are neurologically better suited to bare S‑expressions, which skewed language evolution toward syntactic minimalism, possibly at the cost of broader adoption.

Code Reuse, Portability, and Living Images

  • A major critique: Lisp/Smalltalk “living images” made code sharing and merging hard, especially across vendors and environments.

    • Change sets in Smalltalk and “systems” in Lisp existed, but large teams reported painful merges and clashes, leading to tools like Envy/Developer.

  • Others counter that Common Lisp plus portable UI toolkits (e.g., CLIM, CLUE/CLIO) and NFS‑based file sharing enabled substantial cross‑platform reuse, though vendor fragmentation remained.

  • Contrast with C + Unix/Linux: despite fragmentation, the relatively stable C/ABI model and open‑source culture made reuse and portability far more practical at scale.

Language and Platform Convergence

  • Some see language “convergent evolution”: many modern languages drift toward an OCaml‑like feature set; macro‑heavy languages often “turn into Lisp.”

  • Discussion of Dylan and Apple’s SK8 as examples of powerful, Lisp‑inspired but more readable systems that never reached mainstream despite strong technical merits.

Web, Cloud, and Malleable Systems

  • One view: modern “neo‑mainframes” and complex online systems recreate many lost groupware and multi‑tenant ideas, just via the web and cloud.

  • The web is proposed as today’s most malleable, user‑hackable environment (via userscripts, web components), potentially a path back to “soft,” user‑shaped computing.

  • Cloud vs local is framed as another recurring cycle: centralized systems become complex and expensive, prompting renewed interest in local/self‑hosted alternatives.

Reception of the Article

  • Many readers praise the article’s depth and alternative history framing, despite some simplifications.

  • Skeptics question specific claims (e.g., about modern tooling, GitHub familiarity), but supporters respond with counter‑examples and historical citations.

  • Overall, the thread treats the article as a valuable, if opinionated, lens on how economic, social, and technical forces shaped today’s relatively homogeneous computing landscape.