Dream Dance

Day 868 - The Dangerous Grief of the Mother

It's been windy lately. One might normally hope for a breeze to cool things off, but somehow at 36 degrees and 100 percent humidity a gust of wind feels like being trapped in a blast furnace. It's not a good time to be out.

And yet there's no putting it off. The solar collector has been sending me messages for months, growing slowly in urgency but never rising past the threshold level needed for me to attend to it above my more important tasks. There's always something that needs doing and my scheduling modi estimated that the system could limp by for several more months without attention. Then the wind started. Sudden pressure against the long-neglected gear train ground the whole system to a halt.

Now it's actually worse than the simple cleaning and lubrication that I should have done. I'll have to replace at least one bearing, possibly two. I considered printing a whole new unit so it would be a simple replacement, but it would have taken too long and I would have been out of reserve power before it was finished.

Thus was I halfway through the job, sweating profusely and struggling with the jammed component when I finally noticed Shell watching me.

She was standing at the edge of the clearing, about 20 meters away, very still in the way of Tanzen adults, facing directly towards me. I froze, startled. I'm ashamed to say that my first thought was that I had left my sidearm back at the hab.

As I turned slowly I realized my fear was groundless. She wasn't a hunter, she had no weapon, and if she had meant me harm she wouldn't be watching me out in the open. But why was she there? The village Tanzen had never disturbed me on my hilltop -- in fact the logs show that they actively avoid my hab and the surrounding areas. Sometimes a hunting party will pass relatively close, but since they became aware of my presence here they have ceded this region as mine.

This is to be expected, of course; as groups Tanzen shun interactions with other groups, often going out of their way to prevent accidental contact with out-group individuals. Transients are the exception. Could Shell be turning transient? Possible, I suppose. The recent death of her young son could have triggered changes in her demeanor; I had certainly noticed some. But if she was to go transient she'd be preparing for a long walkabout, not making a four kilometer hike up the hill with nothing but her gathering yoke.

We watched each other for a while, motionless. She's a medium sized female, but she has wide shoulders and long torso with narrow hips, features that males statistically find attractive. She stood motionless, as if captured in a moment of activity, her arms and legs in a posture of action although relaxed and unmoving. The anatomy of Tanzen muscles allows them to hold poses for hours if need be -- even in sleep -- without discomfort. The swaying of the nearby branches in the wind caused the surface of her eyes to shimmer despite her stillness.

I didn't have any such luxury, and my muscles were starting to ache from immobility. Slowly I straightened up and pushed my goggles -- which I had been wearing because of the wind-blown dust and the power tools -- to the top of my head. As I did so she looked up, as if there was something terrible above and behind me. I resisted the urge to whip around and kept my eyes on Shell.

She remained immobile, looking up.

I suddenly felt kind of naked. That's because I was. In the heat I had taken off my shirt and pants and was working in my boots, goggles and bright red underwear. I stooped to gather my tools and gear, all the while keeping my attention mostly focused on Shell. As I started to back away from her I noticed that she had moved again. Her arms and long hands were in a different position than before.

I backed up all the way to the edge of the clearing, and then well into the forest. Then I waited. After enough time had passed that it was clear she wasn't following I turned around and flew back to the hab.

Immediately I pulled up the logs. Turns out Shell had watched me quietly for a good ten minutes before I first noticed her. I let it all play out and then ran the sequence back again. "What the hell was that all about?" I asked aloud.

Day 862 - Measuring Sentience: Cultural Baises

So many tests for sentience assume language ability that I had to do a lot of digging to find alternatives. An old proposal by Koch and Tononi involved just pictures, so it could in theory be applied across a wide spectrum of creatures. The experimental procedure of looking at a set of images and selecting one can be taught by simple conditioning, and then tests with a range of difficulty can be applied to gauge the level of sentience. Tanzen in particular are very visual so this sounds good on its face.

Unfortunately it turns out to be completely useless, for two main reasons. First it's built on a faulty model; and second there are many counter-examples.

The theory was that conscious, sentient creatures process information differently from the non-sentient. In particular, incongruous elements in a visual scene -- while invisible to an unconscious animal -- would stand out to a brain with enough connections to be able to see how otherwise individually reasonable features don't actually fit together.

An interesting idea, but not a strong enough model to distinguish between animals that perform this task reasonably well and actual sentient creatures who don't. For example there are relatively common mental disorders that render humans rather feeble in the face of this kind of visual discrimination task. Obviously being bad at image recognition doesn't make a human non-sentient, and shouldn't disqualify an entire species with similar deficits.

We can't just look at the positive outcomes of the test either. The entire field of computer vision is based on the principle of using contextual data to puzzle out and fit together the components of a scene into a reasonable whole. That's how the vision subsystems of AI programs work. Far from being oblivious to incongruities, vision systems are very good at noticing them.

It's no great surprise that cognitively average humans are also good at this kind of thing; that's why the test is supposedly a measure of conscious intelligence. The problem is that humans do it mostly unconsciously, without control or awareness. Experiments show it, and it makes sense intuitively. We don't get presented with a watercolor pastiche from our retinas which we have to consciously parse to find the important features. No, we see the world first as a complex 3D whole, predigested into its salient parts and attributes before we even get a chance to start to think about it.

The best we get to do consciously is resolve high-level ambiguities. Perhaps this is the idea of the test, but if so it's worthless in yet another way. Making an image that is reasonable on its face but makes no sense in context requires a context -- a cultural basis. If we already understood the cultural basis of Tanzen understanding we wouldn't need to test if they are sentient, and if we don't know it we can't develop the tests to measure it based on this standard.

Once again, without knowing how minds actually work it's very hard to devise tests for them.

Day 856 - Origin of a Scientific Legend

Katherine Ellis was ever a child of the age of interstellar exploration. Born May 3rd, 2224, her first cries heralded the founding of the Quorum Project; her first steps the initial results from the BNEF accelerator; her first sentences the publication of the Pendrell-Hicks-QAI unification paper. By age two and a half loop drive had been discovered and the species forever changed, and Kate was the first of a generation growing up knowing that humanity was -- at long last -- headed for the stars.

She excelled at languages, earning degrees in linguistics and sociology, but also distinguished herself in engineering and computer science. The world changed again while she was working on her doctorate thesis in the field of neuro-cognitive linguistics, when the Astroglide 11 probe returned with unmistakable proof of intelligent life in a nearby solar system. The planet designated A99Hz was inhabited by space-faring creatures quickly named 'Crabs' for their low-slung bodies and sideways quadrupedal scuttle. Ellis immediately shelved her research and petitioned for transfer to the hastily-formed and poorly-named department of Astro-Anthropology.

When she did finally earn her Ph.D. it was for work in exo-linguistics estimating the bounds on Crab social communication bandwidth. (They were surprisingly low.) She spent the next seven years working with the task force designing the experimental mission to A99Hz, and when it came time to select the crew Dr. Ellis was a vigorous, 34 year-old polymath, year-round triathlete, expert on Crabs and the Crab mission. It would have been absurd to leave her behind.

Including her in the expedition turned out to be a masterstroke, although no one could have predicted why. Once the initial survey revealed that the Crabs were not only not at all like us, but unlike anything we'd ever encountered before, everyone had to learn new skills. Her training as a linguist was especially useless for studying creatures with no language, so she had to reinvent herself in a way that could solve a unique and mysterious problem. So she did.

Day 849 - Ending Comes to Bad Ends

Earthdate: July 23, 2315. Friday. It's hard to think about the events of that day; today. What's happening to me, my twin in time, today on Earth, 500 light-years away. My heart aches knowing that I cannot warn her -- warn myself -- of her folly so easily avoided and yet inevitable.

It wasn't even a good party. Just a few friends of friends, and Tele became too engrossed in solving Wei's stupid problem. I didn't have an answer either but it was obviously such a dumb, sophomore-level paradox that it hardly mattered. And yet he kept coming back to it, bringing the conversation back around to discussing the supposed dilemma over and over and over again. The real solution seemed easy enough to me -- "who cares?" would be my answer.

By the time we left I was annoyed with him, and yet I didn't want the evening to end badly. Our history had not been good lately; he'd either been unhappy with me or I'd been unhappy  with him, and I wanted that to change. I wanted us to have a good time, to end on a good note. So I insisted.

I don't even remember the sex very well. Sloppy and perfunctory, it was over before either of us had registered genuine enjoyment. I did it because that was how I had imagined that we would end the evening, and I think he did it because he felt he was doing me a favor. Perhaps a last pity bump before he moved on to pastures greener. His time with the weird e0 girl was over as far as he was concerned. He was bored, and I had to go.

It was at least two days before I realized that my biochips were reading zero-zero. I was busy and that was low on my priority list at that moment. And I thought -- what could be so bad?

Day 840 - Measuring Sentience: The Turing Test

Alan Turing was the first person to formally contemplate the possibility of mechanical sentience: mind as a product of software. To put this accomplishment in context you have to understand that Turing was a computer scientist before there were computers. Four hundred years ago machines for general-purpose computation were rare, and those that existed filled rooms with arcane electrical equipment. Despite these limitations, Turing imagined the virtually limitless machines we have today, and what they might someday be capable of.

Perhaps they might be intelligent, he thought, even as intelligent as a human. Since no one had even the slightest idea of how minds might work, many thought that intelligence was a relatively simple problem. Teach a computer how to learn, they thought, and the rest would happen automatically.

Of course it was not that easy -- it never is -- but Turing formalized an important insight. We don't know how minds work, so we can't say from first principles what systems are intelligent and what are mere machines. But perhaps we know them when we see them. Like art and pornography, human judgement is a fine discriminator that can be brought to bear on the matter of machine sentience.

The test goes like this. The observer and the subject communicate only by simple text. They can talk about anything for any length of time that the observer deems necessary. When ready, the observer decides if the subject is sentient or non-sentient. Repeat and gather statistics as necessary.

So, we've had articulate computers for ages. Does the test work? Well -- yes and no.

Yes, in that the test can statistically distinguish between humans and computers. In the aggregate humans do better than machines in general Turing tests using untrained observers. But also No, in that the results are neither accurate nor precise. In fact by any objective standard the Turing test is a huge failure.

We first have to reject trained observers. People trained in computer science tend to look for certain "tells" that reveal algorithmic reasoning. They tailor their line of questioning in order to detect the ways that computers solve problems that humans don't. This has been an evolving field where AI programmers have tried to work around the kinds of questions that trained observers ask in order produce more prototypically human responses, but this is a trick at heart. It's only a competition between the experts programming the machines and the experts testing the machines -- missing the point of testing if the machines themselves exhibit sentience.

Untrained observers leave us with a different problem. Most of them have grown up talking with computers of one kind or another, and however poorly those machines have fared in real life, people are willing to give them some slack. Thus some computers get judged as human despite obvious flaws. By other criteria no software is yet sentient, so what to make of the false positives?

More troubling still is that many humans, acting as decoy subjects, fail to live up to the high standards expected by observers. It turns out that some observers are unwilling to accept the simple in-artfulness of ordinary human communication and will instead assume that their subject is a machine. Far too often the clumsy phrasing of a human decoy gets them labelled as mechanical, spoiling the whole test.

Ultimately this is not useful for Tanzen because they lack any form of language, let alone a written one. But it does reveal the dangers of asking questions that lack any firm theoretical foundation. How do we tell which alien species are sentient if we can't clearly tell the difference between humans and our programs?

Day 832 - Judgement of the Wolf

The flight home was crowded this time. Lieff and Ramandeep bound for station 6, Gay for 8, Jes for 11, and Gregol going on past my stop at station 9 to station 21. The glider -- already a low-thrust model with a big cabin and huge fans -- spent a good two-thirds of its time cruising at low altitude collecting data. This was a long, long trip full of stops.

As we said goodbye to Jes on the dusty, deserted landing pad, I felt a jolt of discomfort realizing that I would alone with Gregol for most of the next 7 hours. Daytime. As we took off I dug out my most engrossing research papers and prepared for the longest leg of the flight.

Even for an e1 Gregol is a big man. If not for his accident he'd stand nearly 2 meters tall and mass in at well over 100 kg of pure muscle. As it happened he lost his legs spelunking in southern Africa at the age of twelve. Youthful tragedy turned to adult opportunity -- with meat legs in his mass budget he never would have made the first for interstellar travel. As it was he only had to starve himself for months to hit the 50 kg weigh-in at Callisto.

He was wearing his travel legs today -- graceful aluminum swoops laced with thin ropes of pyronol muscles. Now that we were less crowded he had them strapped into the seat next to him and gained himself some knee-room. Shortly after we had taken off for the final leg he began to whistle an off-kilter tune as he worked at his tab. I hate bad whistling.

After a few hours I sat back and tried to drowse in the humming cabin. Just as I was dozing off he tried to make small talk.

There are times I'll respond to discussion, whether that means engaging or deliberately trying to keep myself aloof. This time I was Just damn tired and I don't think I was polite at all.

"What!?" I asked again, starting to be annoyed. I honestly couldn't hear him above the ambient noise of the glider.

"I never asked you," he repeated, raising his voice, "what you thought about Arch's results."

I was sitting in front and to the left of him, so I couldn't see his face. Physics. The QED thing. "It's fantastic," I said, trying to be neutral. "A triumph of theory."

"Yes," he said. He was quiet, almost melancholy. "Absolutely. But don't you have a concern about what it means for the mission?"

Gregol asking my opinion about the mission. That's a first. I didn't know what to say. "What do you mean?" I asked.

"Do you think we should stay?" he asked in reply.

"I don't know," I said. "What do you think?"

He made thoughtful noises and paused, the long pause he uses when he wants to make sure everyone is listening before he dispenses his wisdom. "I'm reserving judgment at this point," he said finally.

The Alpha Wolf of geology, the king of cites, was reserving judgement. What do you make of that?

Day 825 - On a Wing and a Prayer

The Sigma probe package that fell into the inner system toward Portia L2 started out very small. It was a compact and ingenious two tons when it first arrived, and when charged and fueled from the gas giant the projectile weighed in at just over six. After a few years, when the whole thing dropped into orbit it had invested all of that dust, rock and hydrogen into multiplying its fabrication capacity.

The probe was a factory with a brain programmed for science, and there was a planet to explore.

Most of the sensors that the probe delivered into the ecosystem were tiny. They streaked into the atmosphere like meteors in a clump and then broke apart to finish their descent on plastic gossamer wings. The probe's miniature army floated largely unnoticed into the bustling day/night world.

They didn't go entirely unnoticed. Within the first month after deployment the probe detected that a swath of sensors in a small region of the northern rainforest had gone dead. So it sent more. Those also went dead, as did sensors in neighboring areas, their last transmissions suggesting violence at the beak of an otherwise modest black, feathered face.

The probe's cool, mechanical intelligence sensed a challenge, and sent more sensors.

Day 818 - Tripping and Falling, Destination Unknown

Earthdate June 27, 2315. Tuesday. Things changed fast after my talk with Pi. She took over as my adviser within a couple of days. Old Professor yu-Lefevre had no strong objection, or at least he shouldn't have. He was a nice enough old man, but he hadn't managed to do much for me. I told him when we first met that I wanted to work in Dr fo-Weirmen's Crab lab, and two years later he was still telling me that "these things take time."

Yesterday I got a quick note from Pi that I should report to fo-Weirmen's assistant Monday next week for assignment. I have to say I was pretty happy with the change of regime.

There were a few small glitches. My biochips needed replacement, their store of feedstocks fully exhausted. Normally my dorm doctor would arrange a short stay at the campus surgery center and that would be that. But Pi wanted me to do a new, complete physical exam and insisted that the biochips could be handled at the same time. Four to six weeks from now. Until then injections I could do myself would keep the old chips running their basic functions, a thought I didn't relish.

I was admittedly a bit perplexed by the need for a physical, but if She could get me into the Crab lab in five days I was willing to do a lot of strange things.

I didn't realize it at the time, but after weeks of deliberately not getting attached, I found myself -- without ever expecting too -- looking at Tele differently. His oddness became less just interesting and more endearing. His irresponsibility became less just annoying and more something that needed to be managed. His late nights out because less just his business and more our business. In short I was starting to feel like his girlfriend in spite of the explicitly stated nature of our non-relationship. That was why I missed that his feeling about me was changing too. Less just flirty lover and more .. well, a bore.

Day 811 - YC2: Allies

The last day of the conference was all geology, and Gregol -- who had this time allowed his junior contributors to deliver the main body of the talks -- took the anchor leg and spent a good twenty minutes outlining future research. The erosion-subduction zone; variable oxygenic olivines and a long litany of other process juxtapositions that had never been seen before; the Parthenon range; new minerals and their formations such as Frazilgalinite. "I would have called it Tanzen-ite," he joked, gesturing toward me, "but the name was already taken."

Conversation at lunch was upbeat, even enthusiastic at times -- and why not? Ambitious young scientists had the run of a brand new planet, full of wonders and mysteries. I had to get away from that, so I went for a walk.

The six of them were gathered near the cafeteria, but clustered to the side in the relative shade, talking quietly. Chia moved on narrow but clumsy feet, and Mingrui walked nearby in the hunched posture common to all shepards of two year olds. Chia picked something up and handed it to Min, excited by her amazing discovery and said, "Na ca?" Min laughed and smiled and shook her head a little. Chia proudly held up the tiny bottle in her stubby hand and shook it. Someone had been teaching little Chia about salts.

Their sober mood attracted me, somehow. Wald looked sheepish but broke the ice. "We're all with you," he said.

I didn't know what he meant. "What do you mean?" I asked.

"The vote," he said. "We've all decided that we don't belong here. The Tanzen are as sentient as you or I and this is their world." It all flowed out in a rush, as if I was going to interrupt him somehow.

In fact I was mostly baffled. The vote had simply been to extend the mission from eight years to twelve, and I wasn't sure my vote was just for cover. I may think that the extra time is a good idea.

"They are just like us," Susen said, stepping forward. She then told a story of a Tanzen group she had worked near on a field expedition and her belief in their near-humanity. The others joined in with their own tales. Optimistic at first, eventually I dropped my eyes and shook my head. None of it was remotely scientifically useful.

Min, Wald, Susen, Foel, Umam and Won-Yong. Six who voted to keep the mission short because ... why? They don't want to be involved even in a minor extension of the human incursion on this planet, I guess. I think that may be honorable, or perhaps stupid. A longer time here could let me find the data I need to prove sentience, or a longer stay could get the team more comfortable with life here, and less willing to return.

Wald's "we're with you" returned to haunt me. I didn't tell them that I had voted to stay. They thought I had voted against extension. I was their role model. There were also two others who voted for a shorter mission.

I wonder who they were?

Day 809 - There By the Power of Agreement

"No, you're wrong," I said quickly but without conviction. What if she was right? I felt that my immediate peers -- my friends and colleagues -- wouldn't be swayed by archaic stereotypes, but would everyone?

I had been wondering about some of the auditing of Shintus, Walkran, et. al., some of the strange dismissal of my contribution that I couldn't really explain. It just seemed vindictive and without justification. Was this the explanation -- just old-fashioned sexism? My hesitation give Pi her window.

"No, Mädchen, I'm never wrong about this kind of thing. Someone once said that a woman has to be twice as good as a man to get the same respect. That was a long time ago; things are better now. I would estimate that a woman today has to be 30% better than a man. Give or take." She sat back down at her desk and drummed her fingers lazily. "But that 30% is very tricky, and you've been very, very careless."

Of course all that is my personal life, not really a matter (or subject as far as I can recall) of auditing debates. How would this "schema," as she called it, have found its way into the public mind?

As if answering my unasked question, Pi called up a playback with a few quick taps. It was clearly an awkward bit of video, dark, grainy and blurred in wedges from being rectified from the fish-eye distortion of the original data. But I could make out the main character. Paal sat in a bar or restaurant that I didn't know, with other young people who were talking loudly and laughing, drinking red wine. Why he would drink wine and not beer seemed to oddest thing about it at first.

It took me a few seconds to realize that everyone was speaking Spanish and to train my attention to the translation feed. And then I slowly started to understand what I was reading. It's one thing to know that your ex hates you after you break up -- that seems normal enough. It's a whole different thing to find out that your ex hated you while you were together.

It was also quite clear where the schema had originated. The most important part of Paal's speaking tour in South America had apparently been bad-mouthing me.

The painful rock had formed in my throat again, and tears were streaming, unbidden and quietly down my cheeks. I looked down. I knew that if I closed my eyes I would start to sob, so I forced them open.

It seemed like ages before Pi spoke again. "Fortunately," she said in a very soft and reassuring voice, "there is something more important than reputation. Power."

I looked up slowly and met her eyes. "Reputation gets you cites," she said, "and cites get you power, but with power..." Here she smiled, wry and hungry, perhaps the most genuine smile I had ever seen on her. "With power you can make reputations."