Where Google Falls Short

re: Paul Graham Wants To Blow Up Google.

I believe the author of this piece derives an incorrect conclusion from a limited, convoluted use case. I am also baffled and disappointed by the unproductive claim "[...] no one wants to read walls of text.", especially when the use case they chastise of Google is the ability to locate full papers to analyze them. I hope the author considers omitting this statement in a future revision.

The point the author might have made is, no one wants to read a wall of the *wrong* text, of *irrelevant* text, which is laid out in a format unconductive to how we as people effectively read. But these points were not addressed or presented with sufficient research. Here's the response I believe Google deserves:

The problem is not that "Google" doesn't have good design sense.

Let's explore several different interpretations of design and experience:

• The "single input url" is probably the best, most intuitive interface which can possible exist.
• Thousands of users are delighted by their daily doodles.
• Their analytics system is one of the most used, accessible, and powerful on the web.
• Their APIs are almost as powerful as their underlying infrastructure.
• Google maps is perhaps the most thoughtfully designed service I have ever used. Drew Winget feels even more strongly of this than I, and he is an apple user.

In fact, if we want to be really creative with our definition of design, we can say:

• Google's blazing fast response time (and understanding that this is a property user's value) is impeccable design.

Yes, Google knows what user's want. They have the raw data to prove it. And they design for this. I think this is one of the most objective definitions of good design which can exist -- understanding users' priorities, accommodating their behavior, and experimenting with improvements.

So no, I don't believe Google's problem is "design". I think the problem (which Google has only recently started addressing) is their monolithic and antiquated conception of what search is; where computer search ends and human search begins.

If you want to call this design, fine. Everyone is a designer, design can be anything, programmers are designers -- Fine. I get it. I get it, but this type of arbitrary classification is not productive, other than for people to come to philosophical alignment. This specific stance of "everything is art" is passing the buck and allowing the real underlying problems to persist, just like google is doing to users with their incomplete implementation of search.

In a nutshell, Google is great at design, what they've gotten worse at is developing what user's *don't know* they need. Instead, this innovation is being pushed away to the background of Google, to the R&D of ancillary projects, such as glass and brain. No doubt these innovations back propagate and are re-integrated into search, but not nearly as creatively as their potential affords -- more to fuel their future arsenal, to prevent competition with patents, to move metrics and improve efficiency, rather than to incite changes in paradigms within search; the context under which I imagine many of these initial research are founded.

Enough pretext. Let's get to the meet and potatoes:

Google's service to the user typically* stops as soon as they show you result pages; a list of sometimes millions (if not more) results. This is the problem. Once a list of documents is retrieved, the user must sift through them, one by one, to find the desired content. Isn't this the exact point of a search engine? What google needs is not design, it's a search engine which delves deeper to accommodate its user's use cases. And a search engine which is more sensitive to these use cases (think modes). And they're on the right track with Google Now (zero-click results) and using users' unique search history to determine context.

The average use case of Search is not to find a "page" or document. It's to find a specific piece of information. The *edge case* of retrieving an entire document, is you're trying to (as a professional) discover relevance, context, or intent where such may not be clearly encoded (e.g. legalese or research). This use case exists because there professionals (rightfully) don't trust computers algorithms to adequately understand their needs and goals (which is necessarily true so long as computers don't have complete access to our minds).

Even in this edge case, it doesn't change the the reality -- we only want a document because computers are not good enough to do our work for us. What if they were? How would the world wide web look? This is the question I believe we should be asking and where initiatives (with their emphasis on specific niches and domains) have a huge opportunity to disrupt and change the landscape of the web.

I have my own ideas, on www2.

* Google is doing a great job pushing their knowledge graph forward (which I hear will be released in the next few months as an API) and for the past few years has 0 click results via Google Now.