Like Google, Only Much, Much Worse | Hacker News Reader

Like Google, Only Much, Much Worse

slate.com

20 pointssoundsop35 comments

Luckily, Wolfram Alpha has enough funding (and ego) behind it to survive for at least five years. If they throw serious effort into it during all that time, I'll bet we'll end up with something truly impressive (if perhaps not life-changing).

The core flaw with Wolfram Alpha in my opinion is that it relies on curated data. Such data takes time to source, verify, and translate into the necessary target format.

I sincerely hope the same thing for Wolfram Alpha that I do with the Palm Pre: that it'll get the market leader off its ass and spur true competition.

Wrong, the core advantage to WA is that it relies on curated data. That you know it's from a verified source makes a world of a difference. That it can produce actual answers, not just averages or random tidbits, is essential.

Yeah, that's my main concern as well. That kind of curation just doesn't scale. There's a large body of people out there who can be leveraged, who spend their lives curating and optimizing data sources, especially in chemistry and biology.

What Wolfram Alpha needed was to avoid the Spore-style overhype in the days leading up to release. It was nearly inevitable that a product touted as "the Google killer" would get slammed on the rebound, regardless of actual performance. As long as they manage to avoid getting a unit of measure named for them, or returning results about handing hamburgers to crying mimes while the universe abruptly fails to exist, they'll be doing better than some.

Yes, if they'd just described it as a free, online window into Mathematica for ordinary people, everyone would be talking about how cool it was that it could do such a huge range of calculations.

That wouldn't get as much publicity. Most people hate math.

pg submitted Wolfram's announcement of Mathematica 7 a few months ago, and believes in launching early and then iteratively improving the product. I think he might have been sarcastic.

I think Wolfram really does want as many users as possible. This is a type of search engine, not an online version of Mathematica. Students may stay for the step-by-step solutions to many types of math problems, but statistics on how real people use and seek to benefit from such a product are what Wolfram must really be after.

http://news.ycombinator.com/item?id=369056

I wasn't being sarcastic.

What Wolfram Alpha needed was to avoid the Spore-style overhype

To avoid? I'm quite sure he (or his PR firm) has engineered the whole hype thing.

without the hype, wolfram alpha would have been a sidenote in history that noone heard about

Stop comparing Wolfram|Alpha to Google! When did Wolfram ever say Wolfram|Alpha was supposed to be anything like Google?

If anyone comes out with a website that purports to provide quick access to information, it'll inevitably be compared to Google (or Yahoo, or Live Search, or whatever comes next). Slate's comparison is actually even pretty close; the author took the examples provided and tweaked them slightly.

For example, "running 4mph 30 minutes 5'10" 160 lbs 40 year male" provides an answer, but "golfing 4mph 30 minutes 5'10" 160 lbs 40 year male" doesn't. I think that his conclusion was the most interesting part, about "the eternal problem for any wannabe Google competitor[: ...] Google can easily co-opt such improvements -- and suddenly everyone's got a better Google."

Yeah, Google already functions as a basic calculator and unit conversion tool.

Out of curiosity I tried seeing the the calories burnt were any different for a 60 year-old man or 9000 year-old man. Apparently, they aren't.

http://www28.wolframalpha.com/input/?i=running+4mph+30+minut...

I think Wolfram invited it. They had data sources in Mathematica and didn't try to show how that was analogous or mention it in the hyped built up. I think they are hoping that it will be something as important as Google or Wikipedia though different.

Yep, "Google Killer" gets more publicity than "Mathematica + datasets online".

NB: we're still talking about it: the message will reach everyone who might need it, with zilch advertising budget.

BTW: much as I agree it's not a search engine, I just added it as a search engine to FF, for convenience: https://addons.mozilla.org/en-US/firefox/search?q=wolfram (I got the middle one, by phoque)

Read the article. The author compared the two, so it's fair that the poster did too.

And you'd have to be a fanboy or an idiot to NOT compare Google to W|A. They both do the same thing, only W|A doesn't do it very well. I know since I've used them both.

No, that's the point, they don't do the same thing.

They both have a text box where you can enter stuff, and it will return some results. The stuff you're supposed to enter, and the results they return, though, are entirely different.

“It does at first glance look like a normal search engine, but in fact we have 10,000 computer processors behind the scenes working out an exact answer. It will really change the way we get our knowledge, and we hope it will change the way people search online.”

--Conrad Wolfram

"look like a normal search engine" - translation 'it's like google'

"Change the way people search online" - translation 'it's better than google'

I read both those quotes as "different than Google". He's saying as it may look like Google but it's not really.

Come on, how can you say "it will change how people search online" and not expect to be compared with goog. That's just ridiculous.

People will by default compare to the old but that doesn't mean the new and old are the same! After a comparison, they may decide they are the same, or one is better/worse.

The biggest problem for WA is that it is different enough that people used to the Google way of searching will not see its potential and powerful use cases.

The difference is this: In Google, most queries end up with a list of links to pages that Google thinks answer the query. In WA, you get the answer. Of course Google has the "calculator" functionality, and that's the directly analogous product to WA. That's a Google killer feature: I query and get an answer directly without clicking around and hopefully, maybe, potentially finding an answer. Whether that's good or bad is a separate subject.

Another more subtle difference: the query is interpreted differently by the search engines. WA's interpreter is much more powerful in deciding how to interpret the query; it also tells you how it interpreted your query. Google's queries are mostly keywords and phrases (and they've noted that the average number of words in queries is increasing), and fundamentally they're still interpreted as keywords as phrases not the meanings behind them.

How queries are written (keywords vs natural language) and how they are interpreted is a big area of innovation in the search industry. I don't think they've finally solved it, but they're certainly trying hard: Hakia, Powerset, True Knowledge, and now Wolfram Alpha.

Lookit, it's just a calculator. It's not a search engine. The phrase "computational engine" dresses up the former as the latter.

That said, it's a badass calculator and with some guided navigation, it could be really helpful. Mortgage refinancing, crazy sports stats, economic indicators, healthcare costs - anything that bloggers graph, Wolfram|Alpha has the ability to be really, really useful for. Wouldn't you like to tweak Nate Silver's assumptions and see what would happen? Wouldn't you like to be able to see if you could run a regression analysis to outwit PECOTA? You could wonk off with Ezra Klein to see which healthcare cost reduction opportunities would have the biggest impact on the deficit. And that's just off the top of my head.

It's going to be the best new tool for educated and interesting commentary since RedLasso, and no one will DMCA it to death.

it kind of feels to me like this author willingly jumps into the exact same snake pit he's accusing wolfram and other journalists of falling into.

wolfram alpha isn't a google killer. but to expect any small company to somehow or another come up with a fully functional and completely superior-in-every-way product, and to get it right on the first try, is naive at best.

google has been at search for years and years. wolfram might develop into something that gives google a run. but to expect it to be a deux ex machina, catching up to google's more than a decade of development and billions of dollars sunk after only a year or two's worth of development, is fairly stupid.

Wolfram Alpha is much more interesting to me than Google was when it came out (all Google had going for it originally was that it had no ads). Its calculating abilities are also more impressive than Google's Calculator feature has been at any point in time.

Wolfram Alpha will benefit from strategies that every other startup uses--getting as many users as possible, monitoring what users appear to need, and adding algorithms and data to meet the needs of their most popular queries.

I think Wolfram Research is hotter than Google right now.

For some queries, Wolfram Alpha is excellent. As mentioned in other comments, Wolfram's marketing sucks not the search engine.

Like all potentially disruptive products, it meets the needs of a particular niche well, For example, try "population density" in W-A and Google. Be sure to click on the "More" links to expand W-A's results. Amazing.

Over time Google may have a problem. I'm using Twitter for real-time search, W-A where it's strong, and even using Live.com when Google doesn't return meaningful results.

i think they are missing the point. the vision that Wolfram has for this service is probably not found in the product currently. but you can't get very advanced things if

  1. you don't actually shoot for it
  2. you don't have a solid foundation on the basics
  3. you don't have lots of people testing your service

you're not going to get something great by just aiming because the problem space for this kind of thing is massive. by shooting first you can get a better aim

i suspect the hyping for it was more for #3 than people think

from this perspective i find the data argument weird because data is easy. comparing it to Google is also weird because currently Wolfram Alpha is clearly not an internet search engine in any caliber

maybe wolfram wants alpha to do many of the things that we now consider 'googling;' maybe he doesn't. regardless, the author's point stands that alpha doesn't seem to do too many things that are useful to me on a day-to-day or even week-to-week basis. as a biologist, i consider the results of gene searches on alpha to be, if not useless, then at least far inferior to what i get in more specialized databases. do people in other fields also find this to be true of their own searches?

this will be ideal for high school research papers (especially the edward tufte-inspired graphics) and get some new people into mathematica, but beyond that i'm not too excited about it.

Hmm. Try this Google search: "integrate 1/log(x)". Not very useful, huh?

Sorry, but Google and Wolfram Alpha are different beasts. Wolfram Alpha is aimed at a much smaller audience than Google. That doesn't necessarily mean that it can't be just as important one day -- the number crunchers get everything useful done in this society, after all.

On the other hand, try "4.7L/100km to mpg" in both. Alpha understands the question and reformats it nicely but then doesn't calculate the answer. Google just gives the answer.

it works if you match the ratio (i.e. 100km/4.7L : m/g) but it probably should understand that input and maybe have the result as blah^-1

Another thing I noticed about Wolfram Alpha is that it knows almost nothing about fictional characters. "Who is Luke Skywalker?" produces nothing.

That's the most poignant argument for WA that I've seen yet.

Who needs to do "integrate 1/log(x)" in a website?

i've been using it for plotting/calculations while learninating some subjects. i like it because the interface is elegant compared to say my TI-89 and even Mathematica with its ugly square brackets

or this one (the utility is questionable, but it's the kind of derived data that makes WA "computational")

http://www91.wolframalpha.com/input/?i=population+of+India*p...

Reply on news.ycombinator.com