On Monday, May 19, 2014 5:52:35 PM UTC-4, Jason wrote: > > Craig, > > What about computer/automated trading software that currently executes the > majority of stock trades in the world? See: > https://en.wikipedia.org/wiki/High-frequency_trading > > Then there are also those text messaging services where people pay $1 or > more per text message to chat with what they think are humans (but are > actually bots). Not to mention bitcoin mining, and poker bots. >
Sure, but what we see is that automated trading software works well, until it doesn't, and markets crash. If the global economy does collapse, or is gradually collapsing, it may be in no small part due to the cascading effects of high frequency trading. Its risks are no less than any trader who relies on technical analysis, it just happens much faster. As for the other methods you mention, I would expect that very few will have long lasting success. Eventually counter-bots might make bots unprofitable, or they may become illegal. All of these are examples of unsustainable, parasitic mechanisms. What I'm talking about is the opposite - a machine that does not feed on the deception but actually produces an honest value for itself and for civilization. Craig > Jason > > > > On Wed, May 14, 2014 at 1:58 PM, Craig Weinberg > <whats...@gmail.com<javascript:> > > wrote: > >> An interesting little thought experiment to consider: Is there a way to >> create a program or AI moneybot which can figure out how to make more money >> on the internet than it costs? >> >> I see this as a sneaky way to get at the trans-computable nature of >> consciousness as it brings up issues about the ultimate causes of financial >> transactions. As we know, human motives and senses are required to legally >> cause money to change hands. We spend a lot of time developing schemes for >> security that will protect the power of humans to control how their own >> money is spent. Also as we know, the proximate causes of financial >> transaction over the internet are the digital incrementing and decrementing >> of account data. >> >> Even given a souped-up quantum computer which could break every >> encryption and factor, the idea that there could be an algorithm which will >> be able to reliably and legally extract money from the internet forever >> seems fundamentally flawed. We have primitive moneybots already, in the >> form of malware, but releasing malware carries a risk, especially if it is >> successful enough to catch the attention of police. Also, free protection >> against malware tends to spread as fast as the original threat, so that the >> long term prospects seem shaky at best. Finally, even in the case where a >> moneybot happens to be successful, its use would inevitably destroy >> whatever economy that it is introduced into. As the bot’s automatic success >> eclipsed the ebbs and flows of the real life financial risk, there would be >> no way for a market to compete with a sure thing. We’re already seeing this >> happen in the form of automated trades in hedge funds, derivatives, etc, >> but that’s another conversation. >> >> What would it take to write a moneybot that actually *earns* money >> legally without human intervention? Answering this question, if we are >> being honest, is probably a much higher priority for working computer >> scientists than answering the more philosophical questions about Strong AI. >> The question of what a bot would have to say or do on the internet to get >> people to willingly part with their money, and to do so without complaints >> later on, would seem to be infinitely difficult without the bot being able >> to identify personally with human beings living human lives. Modeling only >> the behavior of data being sent and received wouldn’t work because the data >> has no access to the actual experience of a person receiving merchandise.To >> the moneybot, the only difference between successfully selling something >> and failing is that there is no complaints received, not that there was >> nothing that actually existed to sell in the first place. >> >> A moneybot could find, for example, that people spend money at sites like >> Amazon.com, and could create a website that looks and acts like a retail >> site, but there is nothing that the bot could tell the customer to assure >> them that its arbitrarily generated tracking number has caused the delivery >> of the package. There is no way for the program to know whether there is >> really something to deliver or not, and there is no way for the customer to >> ignore the fact that there is nothing delivered. The program can’t >> calculate that the actual Amazon site has a backend fulfillment machine >> which is composed of real manufactured goods, packaging, delivery, etc. The >> bot could conceivably be programmed to understand what such a fulfillment >> enterprise entails, but it has no way to compute the difference between its >> own in silico modeling of an enterprise and the concrete reality that is >> required for people to get boxes on their doorstep. To the bot, financial >> transactions begin and end in the data. All this to say, yet again, that >> the map is not the territory. >> >> Taking this as a metaphor for computationalism in general, our own >> sensory experiences are the brick and mortar presence of the brain’s >> information processing, rather than the neurology of the brain. What is >> literally in the brain cannot, in and of itself, represent that which is >> not literally located in the brain. Internet marketing data can only be >> used to infer what we do and think, it cannot process what we actually >> experience. A computerized salesman faces the insurmountable task of having >> no model for free will. Knowing that financial transactions take place >> under a particular computable criteria does not explain why those >> transactions ultimately exist and how to selectively attract them. The >> probability of success of any given sales approach changes in response to >> unknowable factors which might make a whole class of products or terms >> unpopular overnight. Human agents will change their behavior to suit their >> own preferences rather than to maintain a statistical model of their >> behavior. >> >> Like a shifting antigen disease, the moneybot would have to constantly >> update its offerings to stay ahead of audiences as they grow resistant, not >> only to specific techniques, but to automated money making schemes in >> general. We are seeing this happen now as spam becomes more sophisticated. >> For a while, shotgunning keywords was a popular strategy and the sending of >> garbage emails presumably yielded some benefit for the spammers for a >> while, but these were very easy for end users to spot and avoid opening. As >> end users keep catching on, and spam filters keep catching up, I would >> guess that the spammers are always chasing slimmer and slimmer margins. >> >> To really make a moneybot that harvests cash from the internet legally >> and perpetually, I think that you would have to model the entire universe, >> especially the psychology of each individual person, and their interactions >> with each other. You would have to model all of human history, really, to >> find which trends might repeat at what times. It’s because of the free will >> thing. *Free will is a feeling with teeth. It allows us to bite into the >> world that we perceive in a way that a deterministic algorithm cannot.*Free >> will is motivated by aesthetics, including the aesthetics of function >> and process, which makes it not just non-computational, but >> trans-computational. Computation is a sensory experience, but it is one >> which lowers the aesthetic amplitude in order to extend the reach across >> subjective worlds from the outside in. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Everything List" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to everything-li...@googlegroups.com <javascript:>. >> To post to this group, send email to everyth...@googlegroups.com<javascript:> >> . >> Visit this group at http://groups.google.com/group/everything-list. >> For more options, visit https://groups.google.com/d/optout. >> > > -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/d/optout.
