Deprecated: Function set_magic_quotes_runtime() is deprecated in /home/mwexler/public_html/tp/textpattern/lib/txplib_db.php on line 14
The Net Takeaway: Page 12


Danny Flamberg's Blog
Danny has been marketing for a while, and his articles and work reflect great understanding of data driven marketing.

Eric Peterson the Demystifier
Eric gets metrics, analytics, interactive, and the real world. His advice is worth taking...

Geeking with Greg
Greg Linden created Amazon's recommendation system, so imagine what can write about...

Ned Batchelder's Blog
Ned just finds and writes interesting things. I don't know how he does it.

R at LoyaltyMatrix
Jim Porzak tells of his real-life use of R for marketing analysis.






Trying Feedburner · 06/07/2007 05:58 PM, Trivial

After talking to the Feedburner guys today, I’ve decided to try it out. I’ve redirected my feeds through the Feedburner system, so if you are seeing this in your RSS reader, then it worked properly.

It was more work than it should have been, involving htaccess file edits. Always scary to play with those.


* * *


Clever Clock... · 06/04/2007 11:58 AM, Tech

Seen on Digg: Easiest Clock to Read EVER

But to save you the hassle:
Home of the PolarClock
Fun to watch initial draw as well… Its very heavy on CPU, but I’m sure the author will fix that later…

Direct link to the Flash display:

Its not groundbreaking, just fun in a “cool idea” kinda way.

Others mentioned in the Digg comments:

Fuzzy Time Firefox Add-on gives time as “twenty past twelve”. FuzzyClock is similar for the Mac.
Timeline is a linear time display.
Chrono Lisa Actual clock as well as online version, it uses layered rings in a different way.

Comments? [1]

* * *


X1 says "New business model: Take away features, charge more!" · 06/03/2007 01:05 AM, Tech

In the latest round of X1’s attempt to figure out how to make money while treating users as poorly as possible, they have now taken back everything that they gave away before. This is version 5.6.4, the “stab in the back” version.

You might recall that in X1 starts to open up…, I had started to revisit my dislike of their business practices, and to think that they had finally decided to work with users instead of screw them.

Well, they decided that was a poor approach. The latest is on this forum posting where they link to the comparison.

There are features most users don’t care about now behind the pay wall such as the below:

But here’s the real “screw you”, “bait and switch”, “we still don’t respect our users” set of removals:

So, yet again, the same folks who decided to “phone home” without telling users and didn’t allow them to turn it off, the ones who tried to say that the “free” version was the same as the old paid version without admitting that they removed features like network search, now goes back on what they gave just a few short months ago. At least now they make it clear: you download, you lose features. I guess that’s an improvement.

It used to take a company the size of Microsoft to treat its users this way, but X1 shows that it can be done by companies of any size.

Feel free to let your IT group know that unless they need the shared index of the X1 Enterprise product, they can tell X1 where to shove it and use Copernic Desktop Search for free. Still highly recommended, still in use on my boxes, and still a winner.

(BTW, Copernic has full group policy handling in their free version, at Copernic Network Deployment. Starts to make you wonder just what X1 is trying to pull?)

Look, its their product. They can shoot themselves in the foot or head all they want to, pull all the features, have no free version, whatever. But to play these games with their users is just wrong. Its disrespectful, and it violates trust. This has been a hallmark of their go-to-market strategy since the earliest days, so you think I would learn. I just expect better from the smart guys over there.


* * *


Cool Visualization of the Day... · 05/30/2007 04:35 PM, Analysis

A Periodic Table of Visualization Methods

Very clever!


* * *


Ubuntu 7.04 and No Mouse on Virtual PC · 05/29/2007 09:05 PM, Tech

UPDATE for 7.10: Same issue on 7.10; same workaround applies. See Ubuntu 7.10 and No Mouse on Virtual PC, or just follow the below directions.

UPDATE: 5/29/2007 a workaround from points to which both say to include the kernel parameter i8042.noloop on the boot. This can be done manually each time on the liveCD (Press “f6” for additional options at the boot screen, and hand type it in on that text line) or if you install it, you can manually add this to the boot config file. On my tests with VirtualPC, this allowed the mouse to work as expected! tells how to edit the boot config file.


I’ve been playing with some Linuxes off and on, and you may recall my post PCLinuxOS: Finally a usable Linux… where I told how impressed I am and continue to be with PCLinuxOS.

I’ve now had a chance to play with Mepis which has some really nice touches, but doesn’t look enough like Windows for a smooth transition, and also the lastest version of Ubuntu, 7.04, which looks just like the old one.

But I couldn’t spend too much time with Ubuntu. Why? Because of a bug which affects some users using PS/2 mice… and VirtualPC. Yes, I still use VirtualPC, not VMWare, because I already have images set up and its a pain to switch over (doable, but a pain). I also don’t want to have yet another set of virtual drivers, so I haven’t put in VirtualBox, though it also gets good reviews.

But the issue is that Ubuntu 7.04 does not recognize the mouse or keyboard in VirtualPC as of this posting; you just can’t type anything or use the mouse in the virtual machine (aka, its a “headless box”). This means that its unusable. And so we are stuck for a while.

Why? The beauty of “open sores”, sorry, Open Source. Whenever there’s a bug, it has to compete with people’s time and interest to get it fixed. An issue affecting a Microsoft app running Linux under a Microsoft OS, not surprisingly, is getting little attention from Linux programmers.

I’m not alone in this need; lots of “dabblers” are frustrated that their VirtualPC Ubuntu install is now trashed, and some people on older hardware are also screwed. And we’ve filed bugs about the hardware issue and specifically on VirtualPC:

ps2 mouse does not work in Feisty FINAL VERSION
Mouse, Keyboard disabled in Virtual PC
among others.

And yes, its a kernel problem, which means it affects Red Hat and others. (Note: the issues appears not to affect Qemu/VirtualBox or VMWare or Parallels or other virtual solutions; its just the way that the kernel checks for certain hardware which wasn’t tested on a) enough motherboards and b) in VirtualPC. If you don’t have the specific motherboards or chipsets at issue, or you use another virtualization solution, Ubuntu 7.04 (or Fedora or others) may be fine for you.)

But at the end of the day, bugs which mostly impact MS users just tend to fall to the bottom of the stack. There is no “priority assignment”, process to lobby for bug, nothing to let us know that anyone will fix or how we can draw attention to it; it just sits there, disabling the MS users who want to try out Ubuntu. This, by the way, from the people who complain that they can’t get MS people to switch from Windows to Linux.

With this kind of treatment, I wonder why?

So, feel free to jump on these bugs if you are experiencing these problems; perhaps if enough people post on the forums and comment on these bugs, we can get a fix out which will make it possible to start introducing people to Ubuntu safely via VirtualPC.

Otherwise, Bug #1 in the Ubuntu list, Bug #1, first reported on 2004-08-20 by Mark Shuttleworth: Microsoft has a majority market share will continue to stay on the top.

Comments? [8]

* * *


What value do Agencies provide? · 05/26/2007 02:34 AM, Marketing

One of the most difficult parts of being a part of the shift of media spend from old to new is working with agencies. Their entire business is changing around them, and like the travel agencies during the rise of Expedia et al., they are not learning from the past and adapting very well.

(Note: I’m focusing on large “traditional” ad shops. Direct marketing shops, boutiques, etc. all have their own problems, and I’ll deal with them another time.)

Of course, with advice like this, one can see why its taking them so long.

Scott Donaton is the publisher of AdAge, and so he’s biased. He sings the same song of the agencies since the 30s…

Agencies have always feared disintermediation, but in some segments of the business it will happen. They’ve got to stop the hand-wringing and figure out where they still add value and be confident of their abilities in those areas. Consumer insights. Creative ideas. Media strategies. Marketers will still need those. Yes, even when behavioral targeting and advanced technologies make it possible to serve the right ads to the right audience at the right time. It won’t all come down to technology.

Don’t Get Blinded by Google Envy

Here’s the thing, though. Are those really the areas that agencies should be investing in? After all, it may not all come down to technology, but a whole lot more will than he believes.

In fact, having seen what the tech can do at Yahoo! and Google, I do think that much of the human intelligence that agencies have waved for so long can be replicated and improved by modern approaches. Advertisers letting go of all their “old school” approaches and letting Yahoo! throw all our power behind a campaign can discover entirely new audiences and ways to get them involved in every aspect of a brand, in ways more enduring than TV and more impactful than print, radio, and other mediums.

But there is a huge missing piece, one that I’m surprised wasn’t mentioned by Scott Donaton. And from my vantage point, its the “killer app” where agencies can add value… and also the hardest to defend, of course.

What should they focus on? Emotion. That is the piece that the tech will miss. We can see all the behaviors and the things that emotion drives, we can track conversion and research and all the rest… but we won’t get at the underlying brand attributes in the head past the simple likes and dislikes, the logical decision making, the buy or no-buy calculus.

Agencies, when they work correctly, find ways to evoke emotion and change entire mental sets, not through logic, but through making you feel good about the field of dreams around a brand. We can help create that via the marketing and content we can create at Yahoo!, but we aren’t the experts at what makes the brand tick. We aren’t the experts at converting what the company making the products can pull off on their own (“we can extend the warranty to 90 days from 30!”) to a fictionalized and exaggerated lifestyle aura of the brand. Very, very few companies can do this on their own (Apple is an exception).

Agencies who live the brand will be the ones that survive the fallout. A focus on strategy and execution on the emotion around the brand, and showing how that drives the only behaviors that matter (purchase and proselytizing; purchase alone is too expensive to buy without getting a loyal convert) will be the area that the technology guys will fall flat on.

To do that, don’t you have to have Consumer Insights, Creative, and Media Strategies (to spread the word)? No, you have to be able to use these to work the brand magic, not necessarily own them. They are just steps on the road, but that’s a steep and winding road, and only a few are talented enough to convert the research into an approach, have creative which resonates, and get it into media in a way which attracts attention and effects affect (I couldnt’ resist that last one, its not the correct usage of the words, but it makes me laugh.)

Some agencies are starting to recognize that emotion and brand, even when you are selling via direct “cost per behavior” models. But even the best still think that TV is the center of their brand world, continually quoting that old chestnut: “A banner never made you cry”, usually followed by a reference to Kodak or Hallmark. And until they figure out that they are still trying to treat 1970s media approaches as the way to win in the 2000’s, they will continue to suffer.

So, Scott is right: Google and Yahoo! don’t care what businesses they kill or save, as long as they make money. And there are things the tech can’t do. So figure out what you can do that they can’t (hint: I did that for you) and turn your business that way. Then watch Google and Yahoo! work for you, instead taking work away from you.


* * *


Interesting Visualizations... · 05/15/2007 07:51 PM, Analysis Tech

I think we will see a resurgence in interest in data visualization. It comes around every few years, and I think we’ll see it again soon.

Here are some to get you started:
16 Awesome Data Visualization Tools

From Digg.


* * *


Decision Trees, R, and Sabermetrics · 05/02/2007 11:43 AM, Analysis

The baseball guys LOVE their stats, and stats guys tend to like baseball. They even have their own name for the use of statistics in baseball: Sabermetrics. A very compelling story of how an organization can benefit from the use of statistical measures is Moneyball by Michael Lewis about the rise of the Oakland As.

And Tom Dierickx in his blog Data-For-All shows a good example of analyzing baseball stats in R to answer a specific question. The best part is that its a quick and concise example of how to create and review a regression/decision tree in R.

Predicting April OPS from Spring Training Stats and the followup Predicting April OPS from Spring Training Stats Revisited are both worth a read.

I tend to glaze over the moment I hear the phrase “anything-ball”, so the fact that I enjoyed these shows that even organized team sports can be made interesting with enough ketchup.


* * *


SPSS installed modules · 04/30/2007 10:42 AM, Analysis

Another thing that I always forget. When you are using someone else’s machine with SPSS and something isn’t working, try


to see what modules (SPSS calls then “components”) are installed.

(BTW, just SHOW . gives all sorts of interesting settings, most of which you never mess with, but fun to know that they are there).


* * *


Transpose in SQL · 04/26/2007 01:11 AM, Database

Publishing so I don’t forget:

Assume your data is this:
Year Quarter Amount
2001 1 100
2001 2 115
2001 3 400
2001 4 550
2002 1 200
2002 2 415
2002 3 300
2002 4 750

And you want to wind up with
Year Q1 Q2 Q3 Q4
2001 100 115 400 550
2002 200 415 300 750

Try this:

    MAX(CASE Quarter WHEN 1 THEN Amount ELSE 0 END) AS Q1,
    MAX(CASE Quarter WHEN 2 THEN Amount ELSE 0 END) AS Q2,
    MAX(CASE Quarter WHEN 3 THEN Amount ELSE 0 END) AS Q3,
    MAX(CASE Quarter WHEN 4 THEN Amount ELSE 0 END) AS Q4
 FROM table1

AKA a “pivot” query.

In effect, you are expanding out the rows, 1 additional row per column with just that column’s info, then “compressing” them with the group by. The MAX is just MAXing 1 real value and a bunch of NULLS, so you wind up with that 1 value again.

This is still manual; you could write a script in PL/SQL or TransactSQL or whatever to dynamically build the appropriate query inside the database and execute it into a new table.

See for a slightly fuller explanation of this approach. AskTom, SQLServer Magazine, and other sources have also examined database specific versions of this, but the above gives the basic idea.

(BTW, he also provides an “unpivot” query:


* * *


On a previous episode...

powered by Textpattern 4.0.4 (r1956)