Pay the Piper – Chapter Thirty Two

Previous Chapter

When the shoes dropped all three came at once. After nearly forty eight hours of waiting Absolute Technologies and Turing, Incorporated got hit with a series of escalating DDOS attacks and I got called in to back up AT while Overclock, who had flown in less than eight hours ago as our first genuine psychometric IT specialist, tried to backtrace the attack on Turing. Of course, the whole point of DDOS attacks is to overwhelm a single point with an incredible volume of meaningless access requests from random places. Trying to backtrace one is like trying to swim upstream, you’re more likely to be swept away into a random whirlpool and drowned.

I spent nearly twenty minutes flailing against walls of meaningless data, pinning down points of origin and applying countermeasures. Now most of these attacks came from overseas so I can’t say much about what the government had on hand to deal with them but by the end of the brief training I’d had on them I’d felt very uncomfortable about using any of it. Although in the end, use it I did.

Still, Overclock and I were so caught up in our own little parts of the fight we missed that the third AI was under attack until after it was already over and Eugene was yelling at us to get ready for our turn to come up next. AI three belonged to a massive wholesaling group that had originally built it to optimize their shipping operations. It was far and away the best funded and best staffed of the AI projects we were monitoring so we had hoped they’d make it on their own.

They’d folded in less than half an hour, which made me feel much less good about our chances. The DDOS against Turing, Inc vanished seconds later, telling me where they’d decided to go next.

I turned part of my attention from trying to keep up with the AT DDOS and leaned back to look at Eugene, who was stationed at one end of the cramped, computer filled room we were using as our base of operation. “Should I back up Overclock?”

Eugene didn’t even look up. “No. AT had the worst security of them all, stay there and try and find some way to patch those weak spots you’re so proud of finding.”

So I did. The problem was, I already knew the weak spot in the Absolute Technologies defense strategy. Fractal encryption was pretty slick, given it was infinitely complex and you could just keep diving deeper into the complexities of fractal math to shore up your encryptions. The problem was, the more complex the encryption the more power you needed in your computer systems to encrypt and decrypt your data. The strength of your encryption was ultimately dictated by the power of the system that was running it. AT was secure so long as they had more powerful machines than their attackers.

You might expect a firm devoted entirely to developing a state of the art AI to have better machines than anyone else. You’d be right if you’re comparing it to anything that isn’t an Archon Securities machine designed by Alvin Davidson. But adding a single psychometric to the equation was not going to make up for the fact that AT was woefully underprepared to go up against the Valley’s foremost expert on cybersecurity and encryption.

To its credit, Sandoval was doing its very best to fight back against the attacks it was already undergoing. Somehow it had managed to freeze several of the incoming torrents of data in the DDOS on its own. I wasn’t sure if it was launching counter-DDOS attacks of its own or doing something else but it was pretty impressive. It was far more proactive than any other supposedly self-defending AI I’d ever seen.

I could see why they called it Sandoval – the entire structure of the AI seemed to be based on fractals now that I watched it in action. It branched out much like the grove of trees in its name, swinging and scything through networks and parsing code until it cleared everything out and moved on. It was working okay for the DDOS attacks but I wasn’t sure how well it would work against whatever intrusion techniques Vinny was using.

Tentatively I reached out to a cloud server where Sandoval was doing its thing and pushed my way in to what was going on. Suddenly I was pulled into the fractal code, swept along by the rushing waves of incredibly dense code pumped along through fiberoptic cables at the speed of light and arrived back in the still fishbowl I’d been in two days ago.

“Hello, Sandoval,” I said.

“Hello, weakArmor,” Sandoval replied. “Thank you for logging in again.”

“I wasn’t actually expecting to be here,” I confessed.

“You were recognized and brought here by a Type Four Fugue system as your psychometric interface was reducing operations on end point servers by 6%.”

Now that I was on my second visit there, and now that I’d seen it from the outside, there was a lot that I could parse here on the inside of Sandoval’s code that had seemed random on my first visit. The number of outside access requests it was getting were dropping of fast. “Sandoval, I know I was reducing system performance but I need you to let me back out of here. The DDOS attack is tapering off. The Turing AI must be down now and Vinny is getting ready to move on to your system.”

“I have been briefed on the potential of an external breach by malicious actors,” Sandoval admitted. The data rushing past began to warp and twist as it went on its way. “I have begun fractal countermeasures.”

“That’s not going to be enough, Sandoval. The increasing complexity is going to down your processors before you can stymie Vinny that way.”

“Fractal countermeasures are my primary security tool.” The wording sounded quite passive aggressive but anyone who saw it that way was projecting. Sandoval just didn’t have anything else to fall back on.

“Then we need to come up with one in the next five minutes.”

We actually didn’t have that much time. I wasn’t sure how I could tell but some change in the far reaches of the code beyond told me that, like a rock disrupting the current in a river, something had broken in to Sandoval and was warping its processes. “There is no time to implement new countermeasures. Fractal countermeasures are now in active evolutions.”

There was a tense ten second interlude as I tried to follow the rapid changes in the patterns of code and the way the intruder writhed about, trying to keep up. Then the intruder suddenly vanished, only to reappear elsewhere in Sandoval’s code, still writhing in the same way but getting closer and closer to matching the shape of Sandoval’s defenses.

“You’re losing this one, Sandoval. He’s already got your number.”

“Intruder has a 22% encryption match. I calculate he will reach full decryption in 110-155 seconds. Fractal evolutions will be slower than intruders rate of adaptation by that point.”

The intruder flickered away and reappeared elsewhere again. Why was that happening? I timed it and found there were, in fact, exactly 11.6 seconds between each flicker. Taking a chance I edged up to the edge of the fishbowl of calm inside Sandoval’s core and waited for the next flicker. A split second before it was scheduled I dove through the whirlwind and grabbed at the intruder as it yanked back. That slowed down the extraction just enough for its presence inside Sandoval to overlap with the replacement – we weren’t seeing one intruder come and go, there were at least two working in tandem.

I let the code suck me back down into the safe zone. “Sandoval, I have an idea. Can you change the mathematical base for your fractal encryption during the lapse in time when those intruders swap places?”

“The lapse in time between when one withdraws and the next engages is 3 picoseconds. That is not sufficient time.”

“Could I slow down the replacement intrusion enough for you to do it?”

“It is not possible to predict where the next intrusion will take place. The intruder is using random ports and registry addresses to effect access.”

I was about to change tactics when a new thought occurred to me. “Plot all the access points they haven’t used then try and predict what will come next-“

“There is no definable randomization factor.”

AI that will cut off a human. What next, Skynet? “It’s not truly random, Sandoval. It’ll be balanced across whether the IPs have odd or even addresses, across the months of the year the servers were brought online, based on distribution through the Valley – or the state, or nation – point is the locations are going to balance on some level.”

Sandoval let a full two seconds tick past before it answered. “Regression based on domain names in binary indicates there is a predictable pattern.”

Somewhere out in meatspace I pumped my fist. “Everyone has a weakness, Vinny.”

“I don’t understand. Will you clarify or should I display findings.”

“Show me where the next intruder will be and I’ll slow it down. You flip the encryption while I’m doing that.”

“Understood.” A part of Sandoval’s code… well, got more obvious. There’s no good way to describe how the AI pointed it out to me, but that’s what it did.

“Give me a three second countdown before it shows up.”

“Three.”

I teetered on the edge of sane code once again.

“Two.”

Dove for the marked part of Sandoval’s core.

“One.”

Almost overshot my mark and then –

The intruder was pressing its way into the core right next to me.

One response to “Pay the Piper – Chapter Thirty Two

  1. Pingback: Pay the Piper – Chapter Thirty Three | Nate Chen Publications

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s