Martin only realized they had arrived when, lost in thought, he collided with the Polish researcher, who had suddenly stopped. Martin apologized to Tadeusz, who turned around and smiled at him.
“It is getting better already, isn’t it?”
Martin did not have the heart to shake his head. He tried to say something, but it felt as if his facial muscles were frozen. However, he did feel there was a warm spot, right above his heart.
Once inside, the rest of his body needed a quarter of an hour to reach its normal temperature again. The lab tent and the common tent were well-heated. Stone Aerospace had transported, via ship, a small diesel power plant that now provided the laser in Valkyrie with electricity, and also supplied heaters, computers, and other things. Mr. Stone had greeted him in person right after his arrival, though Martin could barely remember this.
Nevertheless, the trip had been worth his while, at least for the computers. Stone must have invested quite a bit of money for this, he speculated. Martin could have run a 1-cubic-kilometer cell simulation of Antarctic circulation for the coming two weeks. At NASA he would have had to reserve time on a supercomputer first. Currently, Valkyrie was Stone’s only project, and he seemed to be betting the entire company on it. A drilling robot successfully used on Enceladus could also be sold on Earth.
After his body had reached a more bearable temperature, Martin sat down and rolled his office chair a bit closer to the desk adjusting the seat to the right height. If he used the wrong posture, he soon knew it by the pain in his right wrist. He pulled the keyboard a bit closer, stretched his legs, and launched the debugger. I probably stand no chance of succeeding at this task. Valkyrieand its control software have been in development for over 20 years. The various programmers have documented the code very well, indeed. I should congratulate Stone for it the next time I see him, since this is not always the case. Yet software has the natural tendency to become more complex. In the beginning, there is a routine that is supposed to generate a clearly defined result under specific circumstances. The programmer tests the routine under these circumstances. If he is clever, he also checks what will happen under different ones, if he has enough imagination to visualize different circumstances.
No one can foresee the future, though. Three years later, the module might have to work with subroutines that did not exist when it was created. Five years later, the original conditions for which it was written might no longer exist, but as the first programmer had tested it well, no errors appeared even under different requirements—at least not yet. At some point, reality will test the hitherto unknown limits of the programming, and then a crash will occur. Martin was supposed to help ensure this crash did not happen at a depth of over 3,000 meters.
In the case of a short, primitive program, he would have gone through every line of code. He would have checked which command led to which behavior at which time, whether variables were neatly defined, and whether memory was freed up in time. However, for software of this complexity, such an approach was not efficient. It would have taken Martin months to go through tens of thousands of lines of code, and Valkyrie was supposed to start digging into the ice cap tomorrow.
Of course, Stone’s programmers had already run all sorts of tests.
The danger consisted in a kind of tunnel vision. Martin wondered, Would they blindly trust the programs in cases that appeared too trivial to them? For this reason he had brought along his own testing tools. These simulated an actual mission for the Valkyrie software by transmitting data to it via the interfaces defined by the programmers, known as ‘APIs,’ or Application Programming Interfaces. Then Martin could follow the reaction of the software live in the debugger. Working in this so-called sandbox was also faster, since it could try out various scenarios much more rapidly than in real life. He did not have to wait until rear jet 1 had actually started up; he could cancel the test as soon as the correct start command for the jet had been issued.
Martin started his software tests at the critical moments, and thought, what would happen once Valkyrie had finished making its way through the ice? At that point, a number of components had to change their function. The jets no longer could discharge the heated water toward the front for drilling, but now served as the drive. If the command to switch came too late, they would press Valkyrie from below against the ice cap. The drill vehicle, therefore, must recognize exactly when that critical moment occurred. The software also had to take irregularities into account, such as local bubbles in the ice that might briefly give the impression the goal had been reached. Martin systematically changed the input parameters. For the software, this looked as if it was turning hot and then cold, as if Valkyrie was first being crushed by the ice, and then seemed to be swimming in a viscous slush of ice and water. In all cases, the software reacted in an optimal fashion. This did not mean the passengers would have always survived, however. The drill vehicle had been built with certain safety margins, and if these were exceeded, the crew could not be saved. Nevertheless, this reflects the excellent work Stone’s programmers have done, Martin recognized. The software extended the safe area, which had already been defined twice as large as to be expected in reality, by another 20 percent, as it reacted in the right way to compensate. I really will have to congratulate Stone.
Martin worked intensively for two, three, four hours. He was lost in his simulations and anxiously watched when Valkyrie succeeded against the environment—which he had programmed to be particularly aggressive—and when it failed. Therefore, he was all the more shocked when a warm hand touched his shoulder. His body suddenly jerked, and he almost fell off his chair.
A sonorous warm voice with a Southern European accent softly spoke, “Oh, sorry.” Martin quickly got up.
“No, just go on, I did not want to interrupt you.” The woman, maybe in her late forties, as indicated by her laugh lines, was a bit taller than he. She had long, dark hair, full lips, and broad shoulders. Martin lowered his gaze and noticed the name tag sewn to her uniform read Francesca Rossi. He was feeling hot, plus he was angry at himself, but he could not come up with an answer.
“I, uh…”
“It’s okay, just sit down again. I really did not want to disturb you. They told me you were testing Valkyrie, and as it will be launched tomorrow with me inside…”
“You are the pilot?” Martin sat, as he remembered the crew list. He did not know where he had seen her, even though the image was clear in his mind.
“Yes. Though we are going to be more like passengers tomorrow,” replied Francesca, “if I understand the mission description correctly.”
“I… I don’t know. I just arrived today and spent the whole time online in the simulations.”
“It seems our superiors suddenly lost their courage—or is there something else behind the fact they hired you for this task?” Francesca looked at him with genuine interest. He could well understand her curiosity. If I were to be shot into the ice tomorrow, sitting in a large steel tube… Martin did not even want to imagine this scenario.