Christopher Herot's Weblog

Google Chrome Feature of the Day: Each Tab Is A Process

One of the most important features of Google Chrome is under the hood: each tab runs in its own process.  As Scott McCloud describes in his excellent comic book, previous web browsers ran everything in one thread - rendering the page, plug-ins, and javascript.  In the old days before tabs and complicated AJAX applications this wasn't a big deal, but now a problem with one page can make everything in the browser unusable.  This can be a major inconvenience if you happen to be in the middle of editing a long on-line document or involved in an e-commerce transaction.  Also, with the browser becoming the main focus of one's computing activity, it is more common to leave the browser running all day as one visits hundreds of web sites.  With everything running in one process, memory becomes increasingly fragmented until things slow to a crawl.

Chrome puts each tab in its own process and starts a new process each time you go to a new web site. While this approach consumes a little more time and memory at the beginning, it is more efficient in the long run.  It also means that if there are problems in one tab, the rest of them can continue on their merry way, and it's easier to indentify the culprit, such as a misbehaving plug-in, illustrated at right.

As Mark Waks points out in a comment to a previous post, this brings us closer to the day when the browser becomes the computing platform.

Google Chrome Feature of the Day: Nerd Mode

If you do a lot of research on the Web you've probably come to depend in the tab feature of Firefox - now widely copied in other browsers.  But if you open a lot of tabs you've probably also seen your machine slow down to the point it becomes almost unusable.  The Windows Task Manager shows the CPU is close to 100% used, but the only thing you can do is kill the browser and start over.

Now Google Chrome as addressed the problem with its very own task manager.  Just open the "Control the current page" menu and select Developer>Task Manager.  Or hit SHIFT+ESC if you are in a hurry.  The resulting window shows each task and how much memory, CPU, and bandwidth it is using.  There's even a "stats for nerds" that shows the process ID and memory breakdown.

The culprit usually turns out to be the Flash plug-in, which consume more than half of the CPU even when the offending pages are not on screen.  One wonders if this discovery will cause more people to use FlashBlock.

Google Chrome Feature of the Day: Icognito Mode

I've been playing around with Google Chrome. One feature, which was inspired by Safari and is also in IE8, is Incognito mode - a way of opening a new window such that no trace of the sites you visit is left in the history and no cookies are retained. Google puts it own spin on it with the warning they display:


This is a good list. While pretty much everyone is aware that web sites can collect information about their visitors, most don't know that ISPs routinely sell their subscribers' clickstreams to companies doing market research. And of course the most likely spying will come from the person looking over your shoulder.

Google Chrome

Google's new browser, Google Chrome, launches tomorrow (for Windows - Mac and Linux will take a bit longer.)  In the meantime, you can take a look at the clever comic book which describes it.  They've written their own Javascript engine and done some clever stuff to improve browser performance, but my favorite part is how they describe some of the new security features:

  

Google App Engine

Tonight Google announced their answer to Amazon's EC2, S3 and SimpleDB.  Unlike Amazon's offering, which allows you to create a Linux instance and run anything you want, Google App Engine only runs apps written in Python, but in return it will automatically scale instead of making you use the API to start new instances.  There are also APIs for things like email and authentication.

The best part is the price - free.  (At least for now until your app gets large enough).

Thanks to Robert Scoble for streaming the event via Qik, even if Qik did crash and burn, presumably due to the large number of viewers (1722 last time I looked.)

Privacy Lawsuit over Google Street View

Aaron and Christine Boring of Franklin Park, Pennsylvania are suing Google in the Allegheny County court for taking pictures of their home as part of Google Street View.  While the law is quite clear that a photo taken from a public place is fair game, the Borings live on a private road and assert that Google's photo vehicle ignored a sign to that effect.  They are asking for $25,000 in damages for loss of privacy and demanding that Google to remove the photos in question.   Of course by filing the suit they have already attracted far more attention than they ever would have gotten from Street View and a photo and detailed description of their house at 1567 Oakridge Lane is already available at the Allegheny County Web site.  Also, the photos from Street View are all over the web by now.

It appears that Google may have already removed the Boring images, although the house-number algorithm on Google Maps is sufficiently sloppy that it's not clear which photos match which of the five houses on their street.

On the legal side, I am not a lawyer and don't even play one on TV, but the status of a private road is not obvious.  In Massachusetts roads are often signed "Private Way - Dangerous" but that is intended as a warning that the road is not publicly maintained and thus may not be up to the usual standards of maintenance which aren't that great to begin with.  According to a recent article in the Boston Globe, such roads are indeed open for the public to pass.  Often the issue is one of whether there the way has traditionally been open to the public even if it is privately owned.  Perhaps Google will need to put a lawyer in each car as it takes its photographs.

Google's Next Moves

David Evans of Market Platform Dynamics and Thomas Eisenmann of the Harvard Business School gave a talk today on Google's Next Moves that added some depth to the understanding of Google's business model and made some predictions for the future.

David Evans described Googlenomics:

Google's 75% share of the paid search dollar stems from two simple principles:

1. Secure space upon which to sell advertising and drive traffic to said space.
2. Give stuff away to create space and traffic.

Evans pointed out that the Google Terms of Service give Google the right to place ads in its services in the future even in places and ways that they don't do so today.

The technology behind the ad revenue model has three parts:

• Search engine
• Keyword bidding
• Quality Scoring

Prices are set in a Second Price Auction which has been found by economists to generate the maximum return for the seller.  The resulting price depends upon the keyword purchased, for a Luxury Hotel in New York, the price of the number 1 position is $3.89 per click, but a Hot Dog Vendor only has to pay 59 cents.

The goal for Google is to maximize the revenue per search, as given by the following formula:

RPS = APS x CPA x PPC

where

RPS = Revenue per search
APS = Ads displayed per search
CPA = Click-throughs per ad
PPC = Price per click

Maximizing revenue is not simply a matter of getting the highest price for an ad, hence the quality scoring algorithm which will give favored placement to a lower priced ad if it will generate more clicks.

Interesting facts:

• 40% of searches are related to a potential purchase.
• 40% of searches come from partners, e.g. New York Times, that embed the Google search bar.

Google's superior revenue per search comes not only from its superior technology but also from its scale, which allows it to attract more advertisers who will bid on more search terms.  These two factors account for Microsoft's desire to acquire Yahoo.

Thomas Eisenmann talked about the competitive challenges facing Google and its likely actions:

There are three different screens where ads can be placed: TV, Mobile, and PC.  Each of these has a potential competitor who can impede access, such as the cable companies, mobile operators, and Microsoft, respectively.  Google can counter this threat either by "playing nice" through openness initiatives such as OpenSocial and the Open Handset Alliance or it can play tough and enter the market directly, such as acquiring Facebook or purchasing wireless spectrum.

Another strategy is "platform envelopment" - bundling products together in a way that is difficult for competitors to match.   An successful example is Google Base which combines a free product listing, paid search, and a discount on payment processing.  A riskier case is offering free productivity applications such as a word processor, since it's not clear that peoplle are in a buying mood when they are composing a document.

Eisenmann said that Facebook could be a threat to Google if it started pulling more users into its walled garden, although he was not all that sanguine about social networks.

Similarly, Eisenmann thought Android will be a stretch since it is really hard to make a compelling mobile platform.

Over-the-top (OTT) video - creating an alternative video distribution over a commodity IP network - is another way Google can win.  Google can offer the "same head, longer tail" than the cable MSOs, provide better targeted ads and a superior recommendation engine.  Even if Google undercuts the average cost of a cable bundle and pays above the going rate to a content provider there is still enough margin to have a lucrative business.

In the end, Google's biggest problem may be overreach and arrogance.  I remember my days at Lotus where no matter what we did money rained down on us.  It's easy under such circumstances to believe that the success is a function of one's genius, but when the money flow stops it can be quite baffling.

Some of Evans and Eisenmann's predictions:

• If the Microsoft-Yahoo merger doesn't happen, Google will end up with an 80% share of the on-line advertising business.
• Social networks will "blow over."
• Someone needs to come up with the killer app for Social Networks.

Will Google Save Yahoo from being Assimilated?

Opinions Differ on Open Handset Alliance

Is the Open Handset Alliance that Google recently announced the beginning of a new wave of innovation or is it just tilting at windmills?  Andy Lippman made a good case for the former on Tom Ashbrook's NPR talk show, On Point, making the analogy to television business, where TV sets can be connected to any channel, not just the one that supplied the set, enabling lots of companies to become developers of new wireless services.

George Colony took a more pessimistic view at the Future Forward conference last week, saying that Google was going against the fundamental business model of the cellular business.  Forrester CIO George Orlov later elaborated that while T-Mobile has joined the alliance, its coverage isn't as good as Verizon or AT&T.  Unless it could make the trillion dollar investment to expand its network, the majority of customers would continue to patronize the bigger operators, and those operators would have little incentive to open their networks.

While it's too early to call - the first handsets won't be available until next year - I think there are two reasons to be optimistic.  First, T-Mobile (and perhaps Sprint) are hungry enough to try something new and may demonstrate that the market for mobile data will grow much more rapidly when a new wave of innovative applications are allowed to flourish. Selling lots of data that enables third-party applications could be more lucrative than trying to own 100% of the data and applications business.   Secondly, the rest of the world is not like the US.  If other countries continue to pull ahead in wireless applications, it will become obvious that restricting users to a handful of applications determined around a conference table in Bedminster, NJ is inferior to letting the marketplace decide.

OpenSocial details leaking out

The New York Times and TechCrunch both reported more details today on the OpenSocial set of APIs that will unify All Social Networks (except Facebook and MySpace).  What's especially interesting is the list of SNSs that have signed on.   In addition to the usual suspects (Orkut, LinkedIn, hi5, Friendster, Plaxo, Ning) the alliance includes Oracle and Salesforce.com.  While as a developer I look forward to leveraging all these sites through one convenient API I do wonder how Open Social will deal with the varying policies of the member sites.  For instance, LinkedIn has said they will require applications and/or developers to be certified before admitting them to the system.  How will these differences, and the different underlying features be exposed to the end users?