Back

The document follows the story of the development of the first IMP (Interface Message Processor), a refrigerator-sized computer built by Honeywell to BBN’s specifications. The first IMP, known as IMP Number 0, was delivered to BBN’s systems division building to be tested in the IMP room. Ben Barker, a young hardware engineer, was responsible for setting up and debugging the IMP.

However, when Barker tried to load the first diagnostic program into the IMP, he encountered numerous problems. It became clear that Honeywell did not understand BBN’s interface design and had not developed any diagnostics to test the design. This lack of understanding and testing led to the machine being non-functional. Barker, with the help of BBN’s hardware chief, had to figure out how to make the IMP work as intended.

BBN had a close working relationship with Honeywell, who dedicated a special systems crew to work on the BBN contract. Unlike other computer manufacturers, Honeywell was willing to cater to BBN’s special demands for the IMP project. However, their lack of familiarity with BBN’s design resulted in the failure of the prototype IMP.

Barker spent months debugging the machine, working 16 hours a day. He had a slight palsy in his hands, which made the delicate process of re-wiring the IMP even more challenging. Despite the setbacks, BBN’s hardware team continued to work on improving the design and relaying the necessary corrections to Honeywell.

Meanwhile, BBN’s programming team faced their own challenges. They were tasked with designing a dynamic routing system that would allow packets to flow around troubled nodes and links, rather than being held up. Will Crowther, one of the top programmers in the world, came up with a simple and effective dynamic-routing algorithm.

However, Bob Kahn, another key member of the team, had concerns about the flow control system implemented by Crowther. He believed that it could lead to deadlock and loss of packets in certain network conditions. Kahn and Crowther debated the issue, with Kahn pushing for simulations and visualizations to better understand the problems. However, the rest of the team believed that simulations would not help solve the real-world challenges they were facing.

As summer approached, BBN faced another problem - the delayed delivery of the first production IMP from Honeywell. The programming team had started using a simulation program on a lower-grade development machine, but testing on the actual machine was preferred. With the scheduled delivery to California only a few weeks away, the pressure was on BBN to receive the production IMP and debug it in time.

In conclusion, the document highlights the challenges faced by BBN in the development of the first IMP. It showcases the issues with Honeywell’s implementation of BBN’s design, the debugging efforts of the hardware team, and the debates among the programming team regarding flow control. Despite setbacks and disagreements, BBN continued to work towards building a functional IMP system and meeting their commitments to ARPA.

Words: 489