Search Schmerch

Barney Pell is a smart guy, and I’m sure Powerset is a lovely company, but I have no idea anymore what “better” search would mean. I find pretty much everything I want now, and while natural-language processing always sounds great, improvements in how I submit searches do diddly for me.


  1. Funny. I said the same thing when I first heard of Google. You never know what you aren’t finding until you find it.

  2. Chris is right.
    I’ve spent my summer studying this very concept. I’ve watched google/yahoo fight this out on the round-tables, but I’ve really learned from interviewing real users (hang out at Starbucks and ask folks what would make them switch)…
    Here’s what the research is telling me: Users judge search based on three dimensions: relevance, experience, and trust.
    1) Relevance: Ouch… It’s exceptionally difficult to walk a user through the proof that your results are 10x more relevant…
    2) Experience: Tough… Speed and ease are the most articulated dimensions here. Mike Agostino (now at snap) once told me that users can’t really tell differences once app speed is under fractions of a second. I won’t get into ease.
    3- Trust: Hmmmm…. Ask a google user why they think google results are better? 8/10 will say “I trust google… just because” (at least that’s what they told me). Should this make google proud or scared? My belief is that THIS is singularly the greatest opportunity for newcomers in general search – prove to users that they can trust YOU more than they trust the big guys…
    Andy Grove taught me that any product generally needs to be 10x better than entrenched habits in order to cause behavioral change. Try that on the above dimensions… :-)
    -John Anderson

  3. I’m skeptical too.
    But on your “I find pretty much everything I want now” point: if search got a lot better, perhaps you’d “want” to search for a lot of things you wouldn’t even think of searching for now.

  4. Gabe — That point I agree with, but I think that requires a redefinition of search. It’s not about natural language, but rethinking what I’m looking for.
    Put differentialy, I would need to concede, pace U2, that I still haven’t found what I’m looking for — but I just didn’t know it yet. That’s not inconceivable, but that’s different from what’s apparently in the offing here.

  5. Somebody asked me a similar thing just last night – why would any startup want to play in the search domain now that Google is so dominant. Super-easy to answer. People always think they can do it better, and sometimes they indeed manage it. AltaVista worked _perfectly_ well 8-9 years ago. It used all the same query operators (+, -, phrases, etc.) that Google uses today, it offered relevant results, it covered most of the Web, etc.
    But the who young dudes thought they could do a better job, and they did. Maybe it wasn’t that they really had that much better technology (ok, PageRank was novel and it did work), maybe it was their minimalistic main search page, a relief after the busy portal-like page that AltaVista turned into.
    Anyhow, there is always room for better services. might have been the first, but others might actually be better.

  6. Paul – you’ve got some funky PHP errors popping up here, you might want to check it out. It happened when I tried posting after the preview.

  7. Two things that come to mind about making search better:
    1. The dynamics of web and web spam are always changing. Even if search is good now (and searchers are more savvy), spammers and online moneymakers get better, grow and feed like parasites, therefore search need to continually get better at the same speed, if not faster.
    2. Imagine a day when “search” isnt just for websites, but for tangible things. It may just be a thesis, but concepts like futurist Bruce Sterlings are already being explored in some ways. Bruce says:
    “The primary advantage of an Internet of Things is that I no longer inventory my possessions inside my own head. They’re inventoried through an automagical inventory voodoo, work done far beneath my notice by a host of machines. So I no longer to bother to remember where I put things. Or where I found them. Or how much they cost. And so forth. I just ask. Then I am told with instant real-time accuracy.
    “I have an Internet-of-Things with a search engine of things. So I no longer hunt anxiously for my missing shoes in the morning. I just Google them. As long as machines can crunch the complexities, their interfaces make my relationship to objects feel much simpler and more immediate. I am at ease in materiality in a way that people never were before.”