---
title: "Forty-eight kilobytes was a spec."
date: 2026-10-21
url: https://remiam.co.uk/notes/forty-eight-k-was-a-spec
tags: [Gaming, Hardware, Product Design]
read_time_minutes: 4
description: "Remiam's take on the new Spectrum handheld from Retro Games Ltd and Blaze: why 1980s software built for one fixed machine still works, and why pinning the hardware target is still the best decision on an installation."
---

# Forty-eight kilobytes was a spec.

*Published 2026-10-21 · 4 min read · by Liam (Remiam)*

The ZX Spectrum is back this October as a clamshell handheld, 25 games on board for £109.99. The games still hold up for a plain reason: the people who made them knew exactly what machine they were building for. Most of our best installs work the same way.

This month Retro Games Ltd and Blaze ship a handheld version of the ZX Spectrum. It folds shut like a compact, runs on a 4.3-inch screen, carries 25 games including Manic Miner, Head Over Heels and Skool Daze, takes a microSD card for more, and costs £109.99. A sister model does the same job for the Commodore 64.

The hardware is unremarkable on purpose. It emulates a British home computer from 1982, and the interesting part is not the device. It is that software written for a 48K machine more than forty years ago still plays, unchanged, on hardware its authors could not have pictured.

## One machine, every time

A Spectrum game had one target. The memory, the screen, the colour rules and the sound were the same in every house in the country. Nobody tested across screen sizes, operating system versions or a dozen browsers, because there was only ever one of each. The limits were brutal, and they were also complete. You knew precisely what you had.

That certainty is why emulation works so well now. Reproduce the one machine faithfully and every program written for it comes back. Software built against a moving target rarely gets that second life.

> A constraint you can write down is a gift. A target that keeps moving is the expensive kind of freedom.

## What we take from it

We do not write games in Z80 assembly. But the installations and operational systems we build sit much closer to a Spectrum than to a public website. They run on screens, kiosks and controllers that someone chose, bought and bolted down. That means we can pin the target, and we always should.

- Name the exact hardware at the brief. Model, screen, input, operating system version. If it is not written down, it is not a spec.
- Freeze the build image. Auto-updates on a show floor are a moving target you did not ask for.
- Treat the limits as design material. A fixed screen and a known input let you design every state properly, rather than designing for an average device that does not exist.
- Keep the build and its environment together. The Spectrum survived because the machine is documented well enough to rebuild. Your install should be too, so it can be restored in five years without guesswork.

Should clients care about a retro handheld? Not as a purchase. As a reminder, yes. The systems that last are the ones built for a machine you can describe in a paragraph. When we scope an install, the first page of the brief is that paragraph.

## References

1. [Games Asylum, TheSpectrum and TheC64 handhelds announced for October (2026)](https://www.gamesasylum.com/2026/04/30/thespectrum-and-thec64-handhelds-announced-for-october-sporting-a-clamshell-design-and-25-built-in-games/)
2. [Time Extension, The C64 and ZX Spectrum are being reimagined as clamshell handhelds (2026)](https://timeextension.com/news/2026/04/the-c64-and-zx-spectrum-are-being-reimagined-as-nintendo-style-clamshell-handhelds)
3. [Guru3D, TheC64 and Spectrum handhelds bring portable retro computing (2026)](https://guru3d.com/story/thec64-and-spectrum-handhelds-bring-portable-retro-computing-gaming-experience/)