Durell Software

Durell Software Limited began in February 1983, when Robert White — in CRASH’s words — “plonked a 48K Oric and an Epson printer on a desk in Taunton and began writing home computer software.”

What grew out of that desk is, by the numbers, one of the strongest catalogues of the period: 95, 93, 93, 92, 90, 88, 88, 83, across seven years, from a company most histories of the era do not mention. The numbers are on the record; the company is not.

Fast facts

Founded: February 1983. Based: Castle Lodge, Castle Green, Taunton, Somerset TA1 4AB — an attic room in an old building facing Taunton’s castle.
Founder: Robert White.
Programmers named in the press: Mike Richardson (Harrier Attack, Scuba Dive, Combat Lynx, Turbo Esprit); Clive Townsend (Saboteur); Simon Francis (Critical Mass); Ron Jeffs (Oric versions); Nick Wilson (conversions); Dave Cummings (graphics); Stephen Parker, marketing consultant; Phil Dierks and Mike Evis, business software.
Advertising slogan: “Software getting harder…”
CRASH scores, from the issue 48 index: Fat Worm Blows A Sparky 95% (issue 34), Saboteur 93% (issue 24), Thanatos 93% (issue 35), Scuba Dive 92% (issue 2), Critical Mass 90% (issue 23), Combat Lynx 88% (issue 10), Turbo Esprit 88% (issue 28), Saboteur II 83% (issue 41), Deep Strike 79% (issue 36), Sigma Seven 62% (issue 39).

The kitchen units on the ceiling

Robert White’s route into software is the least likely in this Vault, and CRASH got the whole of it out of him in February 1986.

He trained as an art teacher, finished, and worked out there was not much demand. He retrained in quantity surveying and left college with a first-class degree, having met computers on the course. He then applied — “rather cheekily”, says CRASH — to the Oxford Regional Health Authority, which was using a computer-based building design system to model Milton Keynes District General Hospital:

The model of the hospital was a fully accurate three dimensional representation of the building, right down to the very last doorknob and window catch. The whole hospital was being designed on computer — as you can imagine, it was a massive undertaking.

His mixture of art, computer modelling and surveying got him the job. His first task was to find some kitchen work-units in the model, and the list came back with exactly twice as many as the architect knew were needed.

It took me quite a while to make the conceptual jump needed to find them. In the end, they turned out to be on the ceilings.

One architect had entered the units onto the ceilings rather than the floors; a second had later noticed the floors were empty and entered them again. A man whose job was to reconcile a three-dimensional model against physical reality, finding the discrepancy by imagining the room upside down. It is difficult to think of better training for writing games on a machine with 48K.

He left because he was bored:

I really didn’t want to carry on being someone else’s employee and saw there was an opportunity to set up a business of my own producing computer software. So I made the break.

Starting small, deliberately

The founding is unusually well documented, and unusually unromantic.

White and his wife Veronica moved from Oxfordshire to a house near Taunton owned by his mother-in-law — “so in the early days of Durell there wasn’t a mortgage to worry about”. Veronica kept working as a drama teacher. Robert bought an Oric and a printer, and the printer was the hard decision: “the printer was an investment I thought long and hard about”.

Plan A was not to write games. It was to finish an assembler on the Oric, sell it as a utility, and use it to write the games afterwards. That is the correct order of operations and almost nobody followed it.

He then advertised locally for programmers, which is how Mike Richardson and Ron Jeffs arrived. Richardson had left school early with a half-finished Spectrum game — Jungle Trouble, which took the Game of the Month slot in the Living Guide in issue 1 of CRASH. He wrote Harrier Attack next.

Harrier Attack sold nearly a quarter of a million copies worldwide across formats. It also, CRASH notes drily, “attracted a bit of flak for its scenario — it was the time of the Falklands War”.

Training on games that were never released

The most interesting thing Durell did is a hiring practice, and it produced the company’s most famous game by accident.

Clive Townsend had been a Sinclair owner since the ZX81, where he wrote a Tarot card program. He hung around the Durell offices learning machine code, making tea and — CRASH is unkind about this — “wearing his cool-dude sunglasses”. In May 1985 Durell took him on full time and set him to write a Spectrum game called Death Pit, in CRASH’s words “mainly so he could learn how to apply the machine code he’d learnt”.

Death Pit was never released. Halfway through it briefly looked commercial, then didn’t.

Between sessions on it, Townsend was doing graphics for other people and playing with his own at home:

I’m interested in Karate, and designed a Ninja who just ran around in a building on a scrolling screen. I showed what I had done to Robert in the office one day, and he liked the idea — only he wanted the screen to flip rather than scroll.

That became Saboteur, 93%, and one of the founding texts of stealth as a game mechanic. CRASH records the provenance exactly: Saboteur “rose from the ashes of Death Pit, which was cannibalised for the routines it contained.”

A publisher paying a junior a salary to write a game it expected to throw away, and getting a 93% out of the wreckage, is a training budget most companies of the period did not have and did not think to want. Set it beside White’s stated company philosophy, which CRASH squeezed out of him as a one-liner:

Every game we do should really be an improvement for each programmer.

Townsend also insisted his playtesters be credited — “Rich, and Mat the Fat” — so CRASH credited them.

Simon Francis, and the Dragon that got him hired

The second recruitment story is the same shape. Simon Francis met computers via a lab technician’s ZX81 at middle school, learned BASIC on a Pet, and talked his father into a Dragon 32. He wrote Pit Fiend on it, sold to Microdeal’s pocket-money label, and is candid about it:

The only clever thing about Pit Fiend in my opinion was the fact that I managed to get four voices on the Dragon. Otherwise it was a run-of-the-mill game really.

He showed it to Robert White, who “wasn’t interested in Dragon software” — and offered him freelance work anyway, handing him an Amstrad to learn Z80 on. A Galaxians variant and a couple of Amstrad games later, Durell pulled him out of college at the start of 1985 to write Critical Mass, which took 90%.

Francis on his own 90%: “I’m not entirely satisfied with Critical Mass.”

No storyboards, and a business software department

Two structural details worth keeping.

How games got approved. CRASH: “No-one at Durell bothers too much with storyboards. Game ideas are bounced around the office, with everyone chipping in ideas and suggestions until the basic idea becomes a fairly detailed plan. Then it’s a matter of convincing Robert, the Big Cheese, that the idea is worth turning into a game.” A marketing consultant, Stephen Parker, advised on direction, and White reports the advice plainly: “the more High-Tech scenarios involving vehicles and futuristic equipment are what people want.”

How the rent got paid. Durell ran a business software section — Phil Dierks and Mike Evis on an accounting package — and White is explicit about why:

the business software market represents an area of stability to back up the volatile games market. A game tends to have quite a short life nowadays whereas good business software can sell and sell (and sell).

A games publisher hedging its games business with accountancy software, in 1986, having correctly identified that hits do not repeat. Games programmers lent a hand on business routines “more by way of taking a break”.

The cost of getting better

Running through the profile is a note that turns out to be about the whole industry.

Mike Richardson wrote the Amstrad Harrier Attack in a couple of weeks. Scuba Dive took six months. Turbo Esprit took over twelve. He says so ruefully:

Each successive program is taking longer to write. I suppose it’s getting more difficult to keep up with advances in the quality of games.

That is a programmer in 1986 describing, from the inside, the cost curve that would end the bedroom era. Durell’s response was the training pipeline above: if each game costs more, the answer is programmers who get better each time.

CRASH’s own aside on Scuba Dive makes the same point from the reviewing side. It scored 92% in issue 2 — “in those days, the CRASH Smash hadn’t been invented, otherwise Scuba Dive’s 92% overall rating would have made it one of the first Smashes.” Combat Lynx, next, took 88% and “missed the mark by a couple of percentage points”. Then Critical Mass and Saboteur both took Smashes, which is what brought CRASH to Taunton in the first place.

Why it mattered

Durell is the counter-example to almost every generalisation about British 8-bit publishing. It was not in London, Liverpool or Manchester. It did not buy licences. It did not have a famous founder, a scandal, or an advertising campaign anyone remembers — only a slogan, “Software getting harder…”, which turned out to be a promise.

What it had was a method: hire locally, train on projects you can afford to bin, credit the people who did the work, hedge the volatile business with a boring one, and treat each game as the programmer’s next lesson rather than the company’s next hit. On that method a quantity surveyor in Somerset out-scored most of the industry for four years.

An Oric on a desk in Taunton