Wolfram|Alpha as the first “killer app” of NKS

I know why there’s no post here for the 15th yet. Your hands are shaking.

So are mine. I’m so excited.

the distance from springfield, ma to ny, ny is short by over 20 miles
are the mesurament air miles? if so who need it
we need a land distance for travel (i do not own a plane nither 9.9999 % of population)
thanks

I’ve been waiting to test this new intelligent search engine for weeks and unfortunately it will be 1am local time. Guess it will have to wait for tomorrow morning :-)
Regards from France!

preferly a route practical or short

A next step could be to incorporate an universal transaltion function (all the knowledge in all language will be aviable for everyone)

I don’t know if you take requests for the types of data that you work with, but I would love to see in depth information about carbon steel. Everything from the simple data like alloy contents of different steels, to phase diagrams and Ms temperatures.

It would be awesome to be able to go to your site and type in “Ms of 1095 steel”, or “phase diagram of 1095 vs. o1 steel”.

[No website given -- I have many -- you can Google for my real name above and find me easily enough.]

I was delighted *AND* disappointed by NKS and expect the same here.

Prior to reading about NKS, I had been doing research since about 1991 on data compression. I had an epiphany when testing an encryption system I had developed for a Canadian Bank. One of the tests was to run trivial plain text input (the letter ‘A’ repeated a million times, say) through the algorithm and then run a battery of compression algorithms to ensure that the stream was incompressible (hence had gone from low entropy to apparent high entropy). Although not sufficient, I think it is somewhat necessary for anything but the most sophisticated of algorithms (not counting steganography, etc.) to create an incompressible stream from any input.

I realized that I had a stream that was highly incompressible via conventional means and trivially compressible via the decryption half of the algorithm. I reasoned that for at least *some* things that I call ’streams of interest’ (human developed texts, etc) that there would be a similar transformation that would render a much higher rate of data compression than conventional means.

One of my research techniques was to create automata to exhaustively search various ‘vector translations’ to see if conversion from conventional space to the new space would yield a benefit. These were total shots in the dark and none were successful. At the same time I was doing this, I was examining the (relatively) simple mathematics governing these matters. I think initially I was simply trying to get an idea of the probable size of a ‘bit vector’ that would map algorithmically back to ’streams of interest’.

Although the probability of transforming a random million bytes into something compressible is vanishingly small, the input streams ‘of interest’ that would be translated into a limited set of vectors is also vanishingly small compared to the possible inputs. That is, the things we care to compress are decidedly *not* random at all. Ordinary input streams, even if compressed already and subsequently encrypted are still highly ‘reachable’ through transformation whereas the vast majority of theoretically possible streams are not even reachable themselves, let alone compressible.

Like Mr. Wolfram, I also believe that relatively simple algorithms can (and do) generate much of the world’s complexity. In fact, from my research, I believe that the entirety of existence is predicated upon a single simple algorithm. I am not certain what that is, but for those who wish to take a crack at it, I will offer the following heuristics: It involves, e, i and pi. It is symmetrical, but has a mechanism whereby a feedback cycle can break symmetry. Entities arise from nothing because nothing is equal to minus ‘whatever’ added to plus ‘whatever’. Some primitive ‘thing’ must have a way of interacting with other things and one of the primary ways I am convinced they do is such that two very tenuous ‘Planck level objects’ become both the two widely dispersed entities (each of which could cease to exist in the next state change) and a combined object less likely to cease existence, occupying less space and to the extent they manifest as a wave, with a smaller (likely half) wave-length. I do not have nearly enough brain power to crack this nut because although the algorithm will be simple in the sense of being relatively small it will get complex to calculate to the level of things we know (like atoms and their components) and the various known forces.

Part of the reason I believe the above is because I have a sort of notion of ’survival of the fittest’. The universe would take the simplest course first because it is simpler and hence more likely to arise spontaneously and operate more quickly. The simple way would out-compete more complex methods in the race to form the universe.

The reason for going into the above is to express the notion that the history of the universe may be random, but its outcomes are not. That is, randomness is not the thing that primarily governs the apparent complexity we see. Of ‘chance’ and ‘necessity’, I believe necessity completely trumps chance in terms of how things manifest themselves. There is always a chance that two liters of two mixed gases could separate and un-mix. It would violate no fundamental physical law, but it is so small a chance that its existence is moot. When judged against what random chaos would look like, the universe has a surprisingly small amount of apparent entropy. It is astonishingly simple by that measure. My first best bet for a simple outcome would be a simple mechanism to create it.

It would be wonderful to develop a canon and a method of reaching elements in that canon that would allow phenomenal data compression. This would transform computing immediately. A ten-fold increase in levels of data compression would very rapidly increase the capacity of the world’s storage, bandwidth and CPU power. It would place within reach various calculations hitherto impossible.

Something that I find even more amazing and what I really was hoping we would get with a breakthrough in NKS is the notion that since things like Mozart’s music comes from a very narrow vector space, the transformation that clusters his existing music in nearby vector space (a transformed space) would likely contain some music that he did not write, but could have. The same would hold true for medicines, families of algorithms or any other thing that human beings organize in their minds. It is a real long-shot, but there *does* exist spaces where these things exist *and* they are not as far away as a ‘random’ space would be (random ones are unreachable even in theory). They might be reachable and I feel that the probability that they *are* reachable is good. Certainly, the risk/reward ratio is enough that I have devoted a significant amount of time and money to pursuing it.

How NKS, even as a shot-gun approach helps with this is that it helps to define the reachable universe of vector spaces, helps to elucidate which might be of interest (as they map more or less to stuff we know to be interesting), helps to refine our ability to automate research *and* with luck, if we take appropriate aim, it may help us to ‘bootstrap’ our ability to do NKS more efficiently. That is, it may help us to discover heuristics that allow us to reach hitherto unreachable vector spaces — to find ’shortcuts’ to trim the search trees.

I am skeptical that WolframAlpha will be anything but an ‘interesting’ experiment. To the extent that it brings more to the table, I am not expecting huge things. However, I am still hoping for them. If there are some semi-intelligent guts underneath this thing, I think I can come up with some fantastic questions.

I will note that during the course of my research I *did* determine that progress would require large address spaces and lots of storage and processing power. My local system for this research is a lowly AMD four-core box with 8GB of RAM. It is more than I would have dreamed of thirty four years ago, when I started my adventure with computing. Laughably, I had once predicted that we would be able to simulate human intelligence with about 1GB of RAM and 1GHz of CPU and 1TB of storage. I have more than that now and I think my system may be possibly slower and stupider sometimes than my old 8-bit Atari system with 16K of RAM. It certainly takes a lot longer to turn on.

Despite that fact that due to the math of Group Forming Networks (GFN), I realize that today’s computers have no hope of simulating human intelligence, I *do* think that it will be possible to perform something akin to magic within a couple of decades and whatever it is called, something akin to NKS will form part of the solution. There is something fundamentally broken with modern computing apparatus and I think that NKS experiments will help to show a way around it.

I used to think that the ‘novel’ was of necessity outside the realm of the known and required an essentially random search beyond the realms of the known. However, it is apparent to me that the fabric of ‘what is’ is severely constrained and even at the borders of our knowledge, the nature of the fabric does not change much. We *can* incrementally increase knowledge by systematically nibbling at the edge of the unknown. However, it must be constrained by that which is necessary to reduce the search space and unbounded by that which is not necessary to allow the novel to be discovered. This is much more difficult a problem than one might think. That which is necessary is difficult to determine and getting a random heading into an unbounded direction is much more difficult than most people imagine. Even coin flips have some bias and at the levels we are working that tiny bias might be enough to pervert the entire enterprise.

One thing that is hopeful about this WolframAlpha thing is that I believe it addresses a fundamental issue that NKS did not seem to properly address on its own. That is the issue of ‘canon’ to establish what is empirically probable. Arguably, we need to examine everything that can be digitized and sample every transformation possible, including exploring significant depths of promising transformations. This will take people a lot smarter than me and perhaps a collective (including our tools) that is smarter than any person could be.

Whatever happens here, I really like the fact that regardless of the rationale, it is at least partially exploring something that I was setting up to do myself and lacked the resources to do. I still have, until the end of this month access to many terabytes of bandwidth and storage, but I appreciate the difficulty of setting such a thing up to do useful work. I was unable to do so in the year that was available to me. I am hoping that whatever form it takes, they develop a significant ‘canon’ of material on a scale with Google. Even in its rawest form it would be a valuable resource and hopefully they will make it available to other researchers. As an intelligent distillate, it would be incredible and I (perhaps naively) think that a very significant distillate could be stored in a few terabytes — small enough for researchers to use locally.

[No website given -- I have many -- you can Google for my real name above and find me easily enough.]
Oops. At the time of positng my name was above. It should be above *and* below this

Good Luck!

i´m curious . Take care!

Hi,

Does wolfram Alpha have accurate data for the entire world(Details from All countries on all aspects)– or is mostly Just The US? Unfortunately US is a very small part of the world . FOr example if I type distance between Chenai & Madurai — Will it calculate(These are two important cities in the southern part of India.

Am asking this as the Number of internet users from Countries like India, China & many other APAC countries are increasing exponentially in comparisiobn to the US.

Best regards,

Vijay

I hope this will return power of internet to science too, not only for commercial search contents.
Before internet heavy commercialization, scientific knowledge content search was way better.
Wish you success!!!
Best Regards from Bulgaria.

Does anyone know when this application will be available for use in the USA (ie. what time on May 15th, 2009 for EST)?

I’m pretty excited! Looks like Wikipedia times a million. Isn’t technology great?

Interested to see what this will bring to the table. Good luck launching and update us with how the launch went and first steps of Alpha live!

Bob,
Wow! What a post. Even though I feel I was able to comprehend about 60% of your dissertation, you have massaged my brain. Thank you.

Perhaps WolframAlpha will be the leap, and the conduit to further leaps, most everyone has envisioned.

I’m really looking forward to test this out. All the best to Mr. Wolfram and his team, this is truly a piece of art.

~5 hours to live webcast?

Also wish you good luck, especially regarding the load you’ll have to encounter.

greets, stephan@spamschlucker.org

Sciencefiction come true!

Impressive thing.. would become I guess my favorite toy, better than video or computer games.. hope it can help 3rd world people and kids come to science and education!

Good luck
looking forward to medical and humanitarian issues…dont forget resourced limited settings

Felipe

Guys come on now - this is purely PR BS. NKS is cool but alpha’s got little to do with it. Read it cover to cover.

The Wolfram Alpha Blog Page
http://blog.wolframalpha.com/2009/05/15/live-from-champaign/
refuses to show the comments. It links to the webcast which is no longer active.

Patience and humility are virtues that are yours to discover before promoting your achievements.

this is in response to the comment that WolframAlpha should have been named something easier to remember. I think it’s fine but here is a couple of suggestions:

WolframAlamadingdong

or

Wolfmotherramalamadingdong :D

just kidding guys

Best of luck

e

Very nice, interesting mathematical angle, I still need to figure out how to apply this mathematical feature, I wonder about how the user’s location affects the searcher’s answer and also about how it will work on non mathematical subjects.

I asked question about Ukraine and have gotten absolutely wrong information about languages. I know approximately percentage of Ukrainian - Russian speaking population and was surprised to find wrong INFO about this question.Inside the web page http://www81.wolframalpha.com/input/?i=ukraine
one can find that population that speaks Ukrainian language - 97%, but people speaks Russian - 1,7%.
On the other hand, according the Wikipedia http://en.wikipedia.org/wiki/Russian_language_in_Ukraine 34 % of Ukrainian claims that there native language is Russian.
I love this WolframAlpha and enjoy studding new staff with it, but probably after creating successfully working engine it’s time for creating truthful database?

When will I be able to download? Todays date 5-22-09
Thanks,
King

Great post! Just wanted to let you know you have a new subscriber- me!

Just wondering if there is an online virtual keyboard that we can use to input special symbols.
Example) degree - temperature, subscript, pi, etc