On Friday Wolfram hopes to launch WA to the world. My prediction *right now* is that it will flop; allow me to explain why.

Incidentally this is going purely on all I have so so far: I dont actually have access to the Alpha. I could be wrong...
I admit I have been dubious from the start BUT I was willing to believe the marketing hype about WA and assume they really could do what they promised: interpret our human questions accurately and provide  relevant data. Indeed what I have seen so far suggests they have managed to go some way towards that (here is a video of it in action) but they are missing some crucial ideas/points/thoughts.

Humans aren't machines

Notice in the video that he uses a / in the France / Italy GDP question. The engine reconises this as wanting you to *divide* the 2 GDP's. Now I can understand the interpretation, but I dont agree that if your marketing this to lay people it is the right choice. In a normal English question it would be common to use / to mean "I want to see the GDP of France and Italy". See the issue? Dividing the numbers (in this case especially) is unlikely to be what the user is after - and so it fails.

Now the argument back to that is they can use a space, or vs. BUT that is not a good answer. Mostly because it is not natural English (it is CORRECT English, yes, but not natural).  I see this happening a lot for other searches. I mean how is x and * handled? How do you do advanced comparisons (prices of food in x,y,z,etc restaurants) in natural English (in that example it might be "which is the cheapest restaurant on the high street of sleaford?". Humans dont naturally turn that into a mathematical form that WA might be able to comprehend - and as far as I can see it does not handle converting such natural queries well. Whereas Google would probably direct you to so local restaurant review sites WA is going to struggle.

Data engines aren't Search engines

That is a lot of data we saw in that video. And *impressive* amount. But how much of it is every day use? I'm not denying it HAS a use (and WA is going ot be a great research resource) but I dont see the daily value. Even the search for the location (Springfield) was useless unless you want weather data - for the average punter where is the hotels, restaurants, reviews, places to go etc etc. I understand Wolfram says this will be added later, but this takes me to the next point

Hand Curated?

The data is hand curated. I sincerely doubt that you can hand curate enough information to make this usable for things like site seeing (going back to the location example). I would argue that Google has much higher chance of giving you relevant data to a search for "Springfield restaurants" (thougjh perhaps that is a poor example because of the Simpsons) that Wolfram ever could. Actually no I am sure WA could nail restaurants for pretty much everywhere. But that is one portion of the data for locations: Google indexes everything it can find and then makes a rough best guess based on your keywords.  Your guaranteed to find something for any reasonable search linked to a place name. With WA there it just no way they can curate all that data - if nothing else because of the frequency at which it updates!

Giving it all that

Im not meaning to sound anti WA. It does seem to be a nifty piece of kit. But lets forget the claims that it will be our new search engine: there is no physical way we have the technology to organically tailor information to human search queries.I dont know if it is Wolfram hype, Google hype (I suspect this is the one) or Media frenzy that is building Alpha up as a Google competitor - but whoever it is they are wrong. As far as I can see this does not handle data in the way we usually search for it - it does handle mathematical and statistical data in a BETTER way than we might usually search for. This is a Wikipedia competitor - not a Google one, and a strong one at that (because, crucially, it is sourced!)

Lets call this what it is: a curated version of Wikipedia with oodles of extra statistics sat behind a locgic engine combined with a calculator. Advanced yes. Exciting yes. Our new search engine? Nope.