Today’s topic is anticipation and context, two themes which drive a great deal of our human interactions. Even when two people don’t know each other very well, a shared context gives them an enormous head start in mutual understanding. As I started to write this, I carried out a little experiment. I typed into a Google search page “who won the cricket world cup?”, and Google correctly guessed that I was interested in the most recent one – last year – and gave me the right answer.

Then I typed in “where was it played?” – a question that any human conversationalist would recognise as directly referencing the first question. Of course the Google web search engine has no such context, and the first page of suggested links included some pages on table top board games, medieval music, the date when the musical Hair was released from censorship (1968), and a song by the Dave Matthews band. Can you imagine a set of answers like that turning up in a pub conversation about cricket?

Was this a fair test? Clearly not, one might say. Yet it shows how even a hugely successful search engine can be woefully inadequate at recognising context. If we are going to think of software as intelligent, it has to be able to offer suggestions which are contextually appropriate – answers to questions, for sure, but also behavioural changes that adjust to changing situations and an awareness of what I am actually seeking. If something cannot adjust like this, we are unlikely to think of it as intelligent.

Now, all of the major players in the online world are aware of this, especially when the results are being delivered to a mobile phone where my patience level, reliability of connection, and willingness to trawl through dozens of potential matches are all pretty low. We want the system to know enough about us that we don’t have to pedantically explain the same stuff over and over again. Mobile search is going this way, and mobile route planning is already there.

But… the flip side of that is security, or trust, if you prefer. Do you want the major online players to know so much about you that they can anticipate your every whim? Who else gets to know that much about your actions? In the debate about convenience versus privacy, many – perhaps most – people actually want convenience. Privacy is hard work: it is technically difficult and, usually, inconvenient. Lots of people can’t really be bothered, and simply trust that what they are doing is insufficiently interesting to attract the wrong kind of attention.

Which brings us to Far from the Spaceports. Mitnash and Slate are, in part, ethically motivated hackers. They have expertise in the art of cracking into someone else’s code, but they’re doing it in the interest of tackling crime rather than committing it. But it’s a fine line, and some of their actions probably cross a moral line somewhere:

“I just don’t know what the pair of them will get up to once they get in to the system. They’re going to take this a lot further than I would”…

A purist would say that what we planned to do was not exactly legal, any more than the code inspection at the relay buoy was. But then, there were not exactly any laws that applied to this situation…

But hacking aside, one of the most prominent features of the relationship between Mitnash and Slate is their ability to anticipate what each other wants next. For all the differences in their respective hardware systems, as a pair of coworkers, colleagues, and friends, they work on a rather unconscious level with each other.

As an aside, and I’ll be saying more about it later in the week, there’s a Facebook launch event for Far from the Spaceports next Monday, December 7th, 7pm-9pm UK time.

Facebook event banner

Carrying on the series about human-machine relationships, today’s topic is intimacy. I’m not proposing to talk about sex specifically – nor do I in Far from the Spaceports – but a much wider spectrum of close relationships.

In the book, Mitnash has a long term human partner back on Earth. She’s called Shayna, and we only actually meet her in one scene near the start, though she is a regular background presence throughout.

The main relationship that we see is with Slate, his working partner, who also happens to be female gendered. She has no physical form that would distinguish her from any other virtual persona, and with a bit of preparation can adapt herself to a wide range of available hardware.

So their relationship is not on the basis of bodily shape – I didn’t want to write an android book, and the difficulty of getting Slate close enough to the action to be useful is an important narrative ploy. But clearly they are a close-knit couple. As Slate comments to Mitnash about a particular data file she has intercepted,

“there’s actually more about me in the packet than Shayna.”

To which Mitnash replies,

“best not to tell her that, if you don’t mind.”

Their intimacy, the way I see it, rests on two things. Firstly, they share intense and difficult experiences together, supporting one another in them to the best of their ability. But secondly, they communicate with one another in a direct, constant and intense manner. Use of a cochlea implant and subvocal transmitter – originally simply to avoid having to speak out loud in situations where this would be awkward – means that Mitnash communicates not only what he is consciously framing in thoughts, but also a whole other level of half-framed thoughts and ideas.

“Slate, how much do I talk to you without knowing it?”

“All the time, Mit. You murmur to yourself while you’re thinking, and you subvocalise throughout the day. There’s very little about your thought life I don’t know. Or your fantasy life. You’re whispering to me almost all the time.”

“I suppose that means you know all sorts of things I have never told Shayna.”

We are clearly a very long way from this level of artificial intelligence just now. All of the major players in today’s online world have been working on this – Apple’s Siri is probably the best known, but there are many others. At the moment they are all quite gimmicky – after asking Siri what the meaning of life is, and showing your mates that you can send messages and be reminded about events, most people get bored with him (or her in some countries) and the level of interaction drops. Siri and that whole current generation of virtual assistants are just not interesting enough.

Sounding relational, as opposed to encyclopaedic, is a really hard problem in machine intelligence. I think most people remember with dislike Microsoft’s Office Assistant, with its cheerful chatter like, “it looks like you’re writing a letter… can I help?”. I actually thought it was a brave effort back then, but obviously I was in a minority and the whole idea was quietly dropped for another day.

The single best known benchmark for all this is the Turing test – basically you are allowed to chat without being able to see the other person, and have to decide if you are talking with a person or machine… without asking leading questions like “what are you made of?” Part of the test – certainly in the way it is conducted nowadays – is seeing how the entity at the other end deals with abrupt changes of direction in conversation, with ambiguous or poorly defined statements, and with questions where the speaker cannot possibly know the answer.

To date, nothing yet built does very well at the Turing test, despite massive improvements and changes in recent years. As I said, it is a really hard problem, and the numerous “digital assistants” already in use, succeed primarily because they are operating in a very limited domain, with a very constrained set of questions. Do I think we will get there one day? Yes indeed, but I don’t think it will be for a few years yet.

Next time… anticipation and context

I thought for the next few blogs I’d talk a little bit about artificial intelligence, seeing as how the relationship between the human investigator Mitnash and his virtual partner Slate is at the heart of Far from the Spaceports. Quite a few years ago now I used to work in AI, though at the pattern recognition end rather than personality creation.

AI has been a key strand in science fiction for many years, long before it came anywhere near possible in reality, and there is a long history of making entirely wrong guesses. Asimov’s earlier books certainly saw a key role for AI, and his Three Laws of Robotics rapidly became a basis not only for his mobile robots but also as a framework for the way other people thought about AI. But for what you might call serious work, like managing a company or a nation, Asimov was locked into the idea that the machines would be physically huge, filling whole buildings, and would need whole squads of highly specialised operatives to make them work. The concept of virtual environments which were geographically dispersed, like a company network, or indeed the Internet as a whole, escaped him.

Other writers or film makers had different blind spots. One often comes across fictional computers which are able to carry out vastly complicated calculations, analyse and direct the course of spaceships or the economies of worlds – yet output the end results of their deliberations on paper tape. It seems that the hardest things to get right are the interfaces that connect the human and machine worlds.

Often, authors have signalled the presence of artificial intelligence by means of stilted or artificial speech, failing in various ways to match human expectations. The android Data, in Star Trek, could never manage verbal contractions, so always said things like “I can not” rather than “I can’t”. This failure to attain informal speech lasted until the installation of an “emotion chip” which among other things upgraded his language faculties. Apparently verbal contractions are emotional rather than grammatical!

So I wanted to portray the relationship of Mitnash and Slate as one of normal intimacy between friends and coworkers. Each has the advantages and limitations of their particular “physiology”, and hopefully each emerges as a distinct personality. This led to a number of specific choices in the book, a couple of which I want to expand on today.

1. Slate, and the other personas, have a definite gender. Slate happens to be female, while some of the others are male. I’ve left it to readers to decide what this means, since she has no external biological indicators of gender. Some people will like the ascription of gender to machines, and no doubt some will not. There’s a sense in the book that machine gender is a relatively new advancement – Slate describes one particular persona they meet as “male, but only just”.

2. I didn’t want the baggage of clumsy language to get in the way of the relationship. So Slate is chatty, informal, but technically skilled and quirky in the way that a professional human coworker would be. Her communication is not only verbal in the strict sense, but includes a number of nonverbal noises that communicate things like satisfaction, frustration, encouragement etc – again, just like a human colleague does.

In terms of current technology we are a very long way from actually developing a persona like Slate. In recent years there have certainly been substantial breakthroughs in both hardware and software, but nothing I have yet seen persuades me that we are going to see virtual intelligences of this quality in the next decade or so. Within a century, perhaps – though this guess may be as far from the truth as guesses that others have made in the past.

Don’t forget – Far from the Spaceports is now on preorder: follow links to

• Amazon.co.uk
• Amazon.com
• Amazon.in
• and other global Amazon sites – search by name!

Next week… the role of man-machine intimacy…

This is a development on from my earlier blog on this topic which generated a number of interesting comments at different places. The issues which aroused comments were mainly to do with the numbers quoted:

Dawn achieved an acceleration of around 1/100,000 that of Earth’s gravity…I am assuming that advances in technology bring that up to 1/20 gravity, which leads to a top speed around 130,000 m/s.

First, the top speed I was thinking about relates to my Earth-asteroid belt journey, not any sort of absolute maximum. My main concern when I wrote it was that it was a tiny fraction of the speed of light. Even for an Earth-Pluto trip the turnover speed is only about 1% of  light speed – which makes life and calculation a lot simpler. If you were able to carry on with the drive over a period of months or years – say in an attempt to travel to a nearby star – then you would have to take Einstein’s relativity into account.

A brief digression on how rockets work. Basically you propel something in one direction in order to travel in the opposite one – the larger the objects you throw out, the faster you get to go, but on the other hand you run out of fuel quicker. Or, the faster you propel them in that direction, the faster you go, but that needs good engine design. In principle you can throw large objects very slowly, but it would be a frustrating business – much better to throw small objects very fast! The fastest ‘propellant’ would be a beam of light, and the most efficient engine would completely transform some piece of matter into that light energy. The Dawn probe used atoms of xenon as propellant, on the grounds that they are easy to ionise (and hence use as propellant) and are also unreactive while being stored prior to use. The exhaust velocity was in the region of 20,000-50,000 meters per second.

Now, so long as you have fuel, you can go on accelerating, and since your ship is continually getting lighter as you burn fuel, the acceleration tends to increase, unless you throttle back to achieve greater efficiency. The next question is – how long does your fuel last? There’s a trade-off here – if you had a more powerful engine then it would use fuel at a higher rate, but since you get to your destination quicker, you might still come out ahead. When Dawn left Earth orbit, about 1/3 of the mass was fuel (a little over 400kg), but then the engines have been used multiple times to achieve all kinds of exploratory moves.

A solution which is popular in science fiction is to gather your fuel as you go along, typically by sucking up interstellar hydrogen in an arrangement of electromagnetic fields called a Bussard collector – the theoretical science is real, though to date we have not actually built spaceships which use it. Star Trek’s Enterprise claims to be equipped with these, alongside the vastly more exotic warp drive to achieve faster than light travel.

If you had a total conversion engine firing, let’s say, a laser beam from the ship, your fuel cost is very minimal for journeys within the solar system – the Earth-asteroid trip costs you only a few grams of propellant, which presumably you would just carry in your pocket. In Far from the Spaceports I have deliberately not gone to that extreme. Nor, in fact, have I quantified the mass required for these journeys, but have made the assumption that it is a real though not dominant consideration. Mitnash has to purchase “reaction mass” at a spares yard after arrival at his destination St Mary’s, but the design of his ship, the Harbour Porpoise, is not overwhelmed by fuel tanks. Imagine something like a few tens of kilograms of fuel – say the equivalent of a few large suitcases.

Finally, none of these figures take into account getting away from the surface of a “proper” planet like Earth. That is a separate problem, involving high impetus. An ion drive is fantastic at maintaining low impulse for long periods of time, but nearly useless at the high levels of thrust needed for lift-off from a planet. I have assumed that there are regular shuttles of some design which take travellers from Earth’s surface to Low Earth Orbit (broadly speaking at a similar altitude to today’s International Space Station), and that the ion-drive ships take over from orbit on the long haul trips.

The solar system is a big place, and journey time is a big issue with present technology. The Apollo capsules took just over three days to go from the Earth to the Moon, and the several current proposals for sending people to Mars involve months of travel time. The New Horizons probe which recently flew past Pluto took the better part of ten years to arrive there after leaving Earth.

Now, for narrative purposes in Far from the Spaceports, I wanted journeys between planets to last weeks rather than months or years. This gives a reasonable sense of remoteness, without the drawback of having settlements separated so far from each other as to make meaningful interaction virtually impossible.

Almost all contemporary space journeys are based on the principle of critical burn points. The vehicle performs a small number of high energy rocket firing sessions at key stages of the trip, typically at start and end, with smaller mid-journey corrections. The rest of the time is spent in free fall, unpowered. This saves fuel, and is the only feasible journey choice for chemically based rockets.

Here, for fun, is a NASA artist’s impression of the Dawn probe using its ion drive near Ceres.

But this occasional burn pattern also means that speed is limited, especially for human travellers. A person can only withstand acceleration of a few multiples of Earth’s gravity, and the total change in speed is therefore limited. You are stuck at whatever speed your short burns can achieve. Logically, it is far better, once away from the immediate vicinity of Earth, to maintain a low acceleration rate all the time. This doesn’t appear to do much over the course of a few seconds, but the cumulative effect adds up to something quite impressive. There have been experiments with this already, using a technology called the ion drive, of which more later.

For the moment just assume that there is a technology able to drive a spaceship with an acceleration of 1/20 the surface gravity of Earth. That gives the occupants a sense of up and down, and generally makes life easier. The recent film The Martian, following from similar ideas in 2001, used a rotating wheel idea to give a kind of pseudo-gravity to the astronauts, but their engine design was on the current occasional burn pattern. But if you had a drive which could be always on at a low level, then the quickest journey time to some remote point is to accelerate at constant rate to the half-way point, and then flip over and decelerate the rest of the journey. With that drive and flight plan, the average Earth-Mars trip takes about a month (just two weeks at the point where the orbits bring the planets closest together). The trip out to the asteroid belt takes four or five weeks, and a trip to Pluto about 4 months. Nicely in the range of what I want in a story.

What about reality? NASA has trialled ion drives on a couple of small probes, notably the Dawn mission which travelled to two destinations in the asteroid belt – the giant asteroid Vesta and the dwarf planet Ceres. Currently it is still in orbit around Ceres sending back scientific data. Dawn did not burn its ion drive engines full time, but it did trial them for blocks of hundreds of days at a time. Using this regime, Dawn took around four years to reach Vesta, having travelled inwards towards the sun first to acquire a gravity assist from Venus. After some science work at Vesta, Dawn took about two more years to migrate to Ceres. That’s longer than I want for the book, but it’s a whole lot better than just coasting there after a high-energy burn leaving Earth orbit.

In comparative terms, Dawn achieved an acceleration of around 1/100,000 that of Earth’s gravity, which is roughly equivalent to going from stationary to 60 miles per hour in four days (!). Even at such a low value, Dawn currently holds the speed record for spaceships sent out from Earth, with maximum speed around 11,500 meters per second. I am assuming that advances in technology bring that up to 1/20 gravity, which leads to a top speed around 130,000 m/s. I feel those are credible goals for technological advance of the hardware, given that I am also assuming that we have the ability and motivation to establish settlements on various rather inhospitable locations throughout the solar system.

Now, the actual distances between planets vary considerably depending whereabouts in their orbits each of them are – for example the Mars-Ceres trip can take anywhere between 20 and nearly 40 days with this engine. But the main thing, and the one I really wanted, was that you can go places within a month or so.