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 26


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.







Google Desktop Search released · 10/14/2004 03:23 PM, Tech Search

I haven't tried it yet... but some others have.

O'Reilly has a good write up. Also, the incredible Danny Sullivan has a good review here.

The Google Desktop Search indexes the following:

No PDF at this time, nor any mention of an API to add new file types. No searching of compresed files, though that's not so bad. Other good "FAQ" type questions at the Desktop Search Support section on Google.

What's stopping me from jumping into yet another search tool? It has too much emphasis on indexing web behavior. Also, it only tracks web behavior for IE (currently); where's firefox/mozilla/netscape support? Its Windows only (not that I care, but others will). Finally, its GUI is really just a web page, kind of weak compared to X1 or other real GUI tools.

In effect, it seems to act like a proxy, intercepting searches you execute and adding local findings to your results, as well as storing and indexing pages you visit.

Its a good start, beating Yahoo, MSN and AOL's attempt to get search tools out... but as I've said here previously, there's a lot a tool needs to do now to be be competitive, and this is an adequate but not a complete winner of an offering.

More when I play with it...


* * *


KXEN makes OLAP analytical... · 09/29/2004 12:10 PM, Analysis

Interesting read by Bloor Research (UK company) Practice Analyst David Norris entitled OLAP with wings where he points out that "KXEN is becoming the most commonly used OEM data mining component within the Marketing Automation segment" and talks a bit about their IOLAP tool, which adds useful and realistic analysis to what is basically the drilldown spreadsheet approach of most OLAP tools.

KXEN can be found at


* * *


Pay Per Person in Crystal? · 09/28/2004 03:50 PM, Analysis

I am getting a bit sick of abusive licensing. I have griped elsewhere about the sheer audacity of companies who charge by volume of data processed using their software on your systems. A case in point is Pilot Software which makes, among other things, a web site analysis tool. Now, if it were an ASP, and they were using image tags on my site and then processing the data on their expensive servers, then I could see how they would charge me per volume of page views.

Only they have the guts to charge per volume for a tool you run on your own server. That's right: you pay licensing fees for the OS and the database, maintenance for the hardware, even power costs... and then Pilot requires an audit of the data you process through their tool on your own systems and charge you based on that. I find this unbelievable. Should I pay MS for Word by the word? How about SPSS per data point or calculation? What a joke, right?

In other obnoxious things, reporting software Crystal (now part of Business Objects) now believes that its output is still its property, even if its emailed. Ed at Infoworld writes that a reader wrote to him: "It does appear in the Crystal version 9 license that you are limited to 10 users that you can e-mail to for every concurrent access license (CAL) you have."

So, this means that even though a user is not using the product concurrently (that is, they are not generating a report, just reading it), you can only mail to 10 users. Of course, the user was not made aware of this when evaluating the product (similar to how Pilot works, btw: always ask!). Brings up the meaning of the word "use" and "access", words that I thought were already pretty clear. I seem to be pretty wrong, at least legally.

It is amazing how products try to restrict how I use them. If I am using your resources, then you deserve some aspect of pay per use. But if I'm licensing your bits to run on my hardware, you have no business telling me how to run my business. Survey companies try to charge per submission... to surveys running on my hardware. Analytic tool vendors try to charge me different pricing because I analyze other's data instead of my enterprise's.

Let's take the tool analogy one step further. If I buy a hammer for use around the house, I may buy a very affordable one. I can use it for 24/7 construction, but it won't last. I won't pay per use, but it won't scale. Now, I can buy a really nice hammer, a professional grade one. It will cost more, but it will last me until I need new features (handles which absorb vibration, for example). I won't pay per use, because I paid more up front. And I don't get additional features for free, I have to buy a new hammer. And I can give it to a friend, but I can't use it at the same time.

What part of this model doesn't make sense in today's market? Oh, the part I left out about "take every penny you can while you can".

Lifetime value is far more important than short-term earnings at the expense of churn. And the smaller your market (Crystal is big in the reporting space, but a joke compared to MS Office), the mre true this becomes. The value of my continued payment of your licensing fees, recommendations, and public support is far more important than a short time boost in fees for a quarter... because now you have to go acquire new ones to replace the ones you pissed off. Its like someone telling you "You can have $1 doubled every year but you can't spend it til 5 years out, or just take $5 each year and enjoy it". Sure seems smart to let the money grow with happy customers, right?

I understand the future of the pay-per-use media such as digital music and movies. I understand that software companies have to make their money. But restricting use is the best way to start a customer looking for a new solution.

You don't have to be an analyst, or even a marketing expert to understand this.

Takeaway: If you believe your customer is "locked in" and you can institute abusive licenses... that's your first sign that the customer is already accepting bids to replace you.


* * *


A43 is worth a look... · 09/28/2004 12:39 PM, Tech

In the early DOS days, file managers were all the rage. Xtree was the "power user's" tool of choice. However, Norton File Commander had its fans as well. There were a slew of dupes and attempts to mimic these tools for Windows as well, and some are still around, mostly as shareware or commercial ware. I haven't yet found one that is worth paying for, though some people swear by PowerDesk

Anyway, I stumbled on A43 via FileForum and gave it a shot.

It's very low impact: just unzip into a dir and run it. As the author mentions, easy to run off of a USB drive. However, I had trouble figuring out some of the features.

So, to save you some trouble: Note the little buttons at the bottom right?

Click on the "double-headed" one. Bingo! Up pops a slew of useful little options:

Just drag and drop a file (or, for the viewer, a collection of files) into the pane and watch things happen...

In addition, the "unzip" and "zip" buttons are not really buttons as much as drag-n-drop targets as well. Try it once to get used to it. (We (as a gui culture) need better widgets for drag targets). You can go into zip files within the tool as directories as well (sort of like Windows Explorer allows currently).

These are all really handy. In working with the tool, it can be slow to refresh a directory if it has lots of files. In addition, it can be memory pig if you use the image viewer often.

But for the price (free!), its a nice addition to the toolbox. Give it a shot...


* * *


Symantec must be joking... · 09/21/2004 05:35 PM, Tech

I have used Norton Utilities since version 1. I was a PCTools fan back in the day (Central Point Software, anyone?) but after they went under, Norton was the only game in town.

Well, nowadays, they are doing everything they can to reject my money and drive me to competitors... any competitors.

They now have product activation on practically every product. As I've stated here many times, that just doesn't fly with me. Now, let's say I overlooked that and wanted to buy Systemworks 2005.

Not only is the Download price the same as the boxed price, but they will charge me an additional $7 to keep the download available after 60 days.

This is unbelievable to me. The download costs them nothing in real money compared to the boxed product. Bandwidth is just pennies compared to the manufacture of the CD, place in case, make box, pack in box, shrinkwrap or seal, stick in crate, ship to inventory, etc. etc. Then, to charge an additional $7 just to make available the fact that I paid for the product? What possible cost could there be to justify this? What, that row in the database costs $7 a year to keep fresh? (Actually, it must cost $7 for 10 months, since it does recall purchase "for free" for 60 days).

And their pathetic spin around why any of this benefits me? says,
"Product activation provides customers with assurance that they are purchasing authentic software from Symantec, which ensures the high level of quality, reliability, peace-of-mind, and ease-of-use they have come to expect from Symantec products.
By using authentic software, users can protect themselves against the risks associated with the use of pirated or counterfeit software, such as software viruses, inadequate or no product documentation, lack of technical support, software compatibility problems, and ineligibility for product upgrades."

Hmm... I always assumed that the bright yellow box, professionally printed manual, non-cdr cd, and holographic seal assured me that I was getting the buggy, often patched, and slow software that I expect from Symantec. So, I don't really need the product activation to assure me of any of this. In fact, a non-product activated version would more reflect the quality brand ethos: They would understand that I may need to reinstall their product multiple times to fix a machine. After all, that's why I buy it.

Oh, and the $7 fee? shows that this is part of how Digital River helps Symantec screw the user. The people I've met from DR are really nice, so I do sort of wonder who over there came up with this silly idea.

Sure sounds to me like something to benefit Symantec and DR and make my life difficult.

Suggestions from anyone as to a useful utilities package?

More great stuff on the silly activation and DRM issues out there can be found on the fantastic blog of Ed Foster of Infoworld fame. Well worth a read.

Comments? [1]

* * *


Server vs. Client VM · 09/07/2004 01:26 PM, Tech

When one installs the Java runtime from, one gets a JVM ready to roll. However, it turns out that you get the "client" VM. Many of the apps I like to run (and the kind I bet you like to run as well) would benefit from the enhancements of the "server" VM.

However, this is not installed by default from the site, though some apps ship with it. Where would one get it?

It turns out to be part of the JDK, the Java Development Kit. This is also called the SDK, and is available at the J2SE site. Note that depending on your app, you may also want to play with the J2EE SDK.

What is the difference? Well, the Server JVM takes longer to start up, and it takes much more memory, both initially and during program runs. The advantages, however, are that you get more runtime optimization. Since CPU and memory are relatively cheap these days, it probably makes sense to run all your java "tools" using the Server JVM.

How to make sure you are using it? Well, one way is to run "java -server -jar app.jar" or whatever; "-client" will run the default client one. However, if you want to change the default from client to server, you can edit the "jvm.cfg" file. This is usually at
C:\Program Files\Java\j2reXXX\lib\i386 on Windows, and
/usr/java/j2reXXX/lib/i386/ on Linux. Usually it will have two lines:
-client KNOWN
-server KNOWN
Swap the order and resave.
-server KNOWN
-client KNOWN

Do I already have the Server VM? One way to test is to just run "java -server". My test machine said "Error: no `server' JVM at `C:\Program Files\Java\j2re1.4.2_04\bin\server\jvm.dll'"

How to install it? Simply make a jre\bin\server\directory, and copy the correct jvm.dll into it. You may also have to make the config file changes above. Or, install the whole JDK and that should give you access to it as well.

If you want the techie details around why the Server side is so different, check out this Hotspot White Paper

Finall, some other tips for making the JVM stuff run more smoothly.

In the old days, you would change autoexec.bat to change the path and add environment variables. Nowadays, you right click on My Computer, Properties, and then Advanced, Environment Variables. You can still muck with the autoexec.bat... but I wouldn't.


* * *


Wikis and Blogs, Drupal and SnipSnap... · 09/03/2004 07:14 PM, Tech

I really hope that Drupal figures out how to marry the wiki ease of editing with the necessary features for blogging. It really is a nice tool. That being said, lots of internal strife recently over the classic "hack some features to meet immediate needs" vs. "good code is hard to write, but it lasts". James Seng has created Drupal4Bloggers which adds many of the MT features to Drupal.

This is very nice... but the mainline Drupal folks aren't accepting his patches fast enough and/or they aren't as well written as they expect and/or the moon is in a certain aspect, but whatever it is, there is talk that Mr. Seng will just keep going with his heavily modded Drupal on his own, known as a "fork" (my opinion, not that of the Drupal folks, who will think I am making a mountain out of a speck). In my experience, forks kill projects, so I would rather not see that happen. Instead, I hope all will continue to play nice, and work to integrate Mr. Seng's features into the mainline Drupal system.

Why mention any of this at all? Because internally, rather than setup either a LAMP box or Apache x%x% PHP on Windows in order to run Drupal or a PHP oriented wiki, I have installed SnipSnap. It elegantly combines aspects of Wiki and Blogs, and is a self-contained Java web server/system. It can use a variety of databases, but includes McKoi embedded as well as simple file system... And get this: You can test it with Java Web Start. Open source, based on the Vanilla wiki, and some pretty dedicated programmers behind it.

This is great for running on a windows box in a corporate org. BTW, I tried the powerful Twiki and it was a nightmare on Windows and IIS. No matter what they say, that's a unix tool through and throug. And I know, I could just install Knoppix or any of the user-friendly distros and be done with it, but SnipSnap was quick and easy. 3 clicks and it was up and running. Very nice.

Finding an affordable external host who will support Java is painful, so LAMP tools like Textpattern and PHP-Wiki and, yes, Drupal will continue to rule the roost for the "public" web. Its interesting: The very things that made me choose SnipSnap are impossible to find externally (_cheap_ hosting with a Java runtime and availabilty of jboss/tomcat in the future), and like many organizations, all the wonderful PHP tools I admire are kind of a pain to get running in a Windows-centric group. Yes, with enough time, I could get Apache/Perl/PHP/etc. emulated or run (Cygwin / UnxUtils / ATT's UWin / GnuWin32 / MS Services for Unix / Countless "all in one AMP for Windows installers") on Windows... but why should I have to when SnipSnap is there? That's how to package software to invade corporate America with open source.

Note to the open-source world: Find easier and better ways to package PHP, MySQL, and Perl with built in webservers not requiring IIS, or provide step by steps to use with IIS and don't require Apache. Make your apps standalone, and provide migration paths. Provide the features you want, but also the features your users need across their unique situations (corporations, personal sites, family photo albums, etc.) Assume that external hosting on LAMP will be different from Intranet hosting on WinXP and Win2K boxes.

Otherwise, you are making wonderful software that that vast majority of "knowledge workers" will never take advantage of.


* * *


Jedit right mouse menu addition... · 08/31/2004 10:13 PM, Tech

Just so I don't forget:

Put the following into an .reg file and import
it into the registry:

Windows Registry Editor Version 5.00

[HKEY_CLASSES_ROOT\*\Shell\open with jEdit]

[HKEY_CLASSES_ROOT\*\Shell\open with jEdit\command]
="c:\windows\system32\javaw.exe -Xmx128m -jar d:\Programme\jEdit\jEdit.jar -reuseview "%1""

you might want to adjust the title (open with jEdit), or have to adjust
the path to javaw.exe and/or jEdit.jar.

The line starting with ' =' is a single line.


* * *


SPSS 13 and phoning home... · 08/31/2004 05:04 PM, Analysis

Ok, I am in a tough spot. I was a beta tester for SPSS 13, and I got to play with some of the new features, including the great classification trees add on feature (no, it doesn’t replace AnswerTree, but certainly reduces the need for it in many cases). Graphics are improved again, and this seems to be where lots of effort was spent. Some nice things here.

Fields can now contain up to 32k of text, meaning long comments or text fields can be kept in the same file with the user’s other responses. A Time and Date Wizard makes date manipulation easier (though they still never seem to have formats to easily read unix log file timestamps, and probably should allow custom formats for imports). Aggregates can now be kept in the same file as the source, ala SAS. Speaking of which, SPSS 13 can read (and write) SAS 9 files. The rest are incremental improvements, in my opinion. Read more about all the new features here.

All good news… but there is one new “feature” I am not ecstatic about. In fact, it is causing me to rethink whether I want to upgrade, or recommend the upgrade, at all.

Now, I don’t know if the released code does this… but the beta included a Registration Requirement. Like Windows XP and the other recent MS products, SPSS 13 generates a hardware-based system ID key and sends it to SPSS to authorize the install.

I fully understand that I do not buy software, I license the right to use it. And, according to the current shrink-wrap license controversies, companies appear to have the right to define how and where I use their software. So, its not like they (SPSS) are doing anything illegal.

But I don’t buy products which phone home. Its kept me from buying or recommending the X1 search tool, it kept me from buying TurboTax, and I use Windows XP only because I had to (I’m not good enough with Linux to do what I want, and I use a corporate edition at work which didn’t phone home.) I don’t buy time-locked media, and I don’t buy media which requires a code to work. The few tunes I’ve bought, I’ve bought via Apple iTunes (more on this below), because they have the least restrictive procedures. If I spend my money instead of pirating a product, I expect to have at least as much freedom of use by being legal as I would stealing it. Unfortunately, yet again, that will not be the case.

I sent notes about this concern to the SPSS gang during the beta, and they were very kind to answer my concerns, including:

Why should this bother me? If I’m not guilty, I shouldn’t worry, right? Well, I am afraid that I am guilty. SPSS can be very slow on large datasets. So, yes, sometimes I will pull 3 old boxes off the shelf, install an imaged OS and install SPSS, and split the dataset. I’ll kick off the same transforms and such on each of the boxes for each part of the data, and magically, my process takes 1/3 the time. I then blow away the drive and put the machines back on the shelf.

Is this legal? Well, it is only one user (me with a KVM). From the old “book” model, its is violating the “code is running on one machine at a time” approach, but then again, that license model hasn’t been used by companies since the Borland days. But I am not a lawyer (IANAL), so, probably yes, I have violated my license in some way by doing this.

Have I duped the CD and shared my license code? Nope. Have I made a “keygen” and posted on 0-day warez sites? No, of course not. But I am still having to suffer the same “install limitation” that the pirates face.

But look at Apple. The iTunes music store allows a purchased song to be burned 10 times to CD. That’s still nothing to the pirate, who needs thousands of copies… but that’s great for me; I can make 3 mix CDs (one for morning, noon and night) and still have room to reburn if I lose one.

Could SPSS do this? Would 10 installs be out of whack? Think of the pirate: They can release their version and license code, but only the first 10 people benefit. Unless they crack it, which would lead one to wonder about why even have the install check in the first place. So, 1 or 2 installs seems absurdly small, just saying “we can, so we do”.

Allowing 10 gives elbow room, while doing nothing to encourage piracy. This is the important point. It allows legal customers to install as many times as reasonable, same as the current scheme expects… but stops broad scale piracy.

After all, that’s why they are doing this, right? It couldn’t be that they think they are losing money on the fact that I install on 3 machines a few times a year (and erase them afterwards), or that a person might have installed SPSS on a contractor’s machine for a month for some extra help. If they really think that the customers are the problem, then they may have to rethink.

From what I recall, its the customers who have kept SPSS in business for 13 versions, even without install checkers. The pirates are the problem, right? Let’s make sure we all keep that straight. Find a way to be really nice to the customers, and let the pirates suffer. They are not the same people for the most part, so treating them all under one umbrella isn’t fair, just, or equitable. All it is is bad business.

(Note: I haven’t seen 13 gold and shipping as of publish time, so none of this may apply. Let me know in the comments. If I am incorrect about the phoning home, then I’ll publish an update here in a new entry and in an update to this one.)

Comments? [2]

* * *


Furl and Bookmarks... · 08/28/2004 01:07 PM, Reference Search

I’ve been a long time user of Blinkpro, a commercial bookmark manager. I have a LOT of bookmarks. How many? Well, I haven’t counted them in a while… but my exported bookmarks.html file is now 2.6 megs.

So, I’ve been very bummed that Blinkpro has started to muck up a bit. I understand that the owner, the smart Ari Paparo (and creator of the recently popular NYWiki) has a day job and a couple of cool hobbies, so I get that Blinkpro may not be top of his list…

But recent errors have led me to look at other options for storing my knowledge. While I’m willing to pay for the solution (I pay BlinkPro some bucks a month), I had some requirements:

Stuff I don’t care about that everyone else does:

I’ll get one of my biases out right now: I think the “community” aspect of these things is kind of a joke: why? People don’t organize themselves, so you wind up just having lists of thousands of people and their bookmarks. Sure, most popular ones float to the top, but then what…? Clay Shirky (of course) loves this aspect of things, read all about it where he also lists a few sites that I haven’t gotten to yet. But watch, if these sites don’t orient their research aspects, they will all disappear. Again, personal bias, you may love the social part of this, so don’t ignore it.

Also, Blinkpro is a pretty tough act to follow. For all my complaints, I push it pretty hard and it does a nice job. So, if you are looking for an online bookmark manager, and are willing to accept that search is good but not great, Blinkpro is a good place to start.

Besides 3rd party sites, I there are a couple of open-source “DIY” options. So, here’s what I’ve seen so far:

Darling of the blog set, check it out here. Web site, so usable across machines. Clever idea: Pulls in the content of the bookmarked page and allows you to search that as well. Nice touch.


Open-source project, available here. Most full featured of the open source projects, this now includes an address book and calendar, 2 things that have no business being in a bookmark manager. Like most open-source projects, little to no docs, too many little detail features, not enough power features. Remember, you have to run this on your own hosting.


Open-source project, available here. Emphasis on team/group sharing, and on being in the sidebar. This becomes annoying fast; every screen is 160 pixels wide. That being said, most refined UI features, including DragnDrop and right-click menus. Remember, you have to run this on your own hosting.


So, what’s the winner? Well, if Furl can get their display together, it will wind up being the most powerful. But currently, hard to settle on it for real work when its just a long list. Yes, good searching, but still… So, playing with B4U and Sitebar, but still relying on BlinkPro (bad searching and all) with more Furl dabbling.

(Also beyond just bookmarks, there are often other knowledge bits one wants to store. I’ve tried Wikis (phpwiki, tiki), CMS systems (Drupal, Mambo), Mind-maps (most recently, TheBrain) and now Textpattern. They all still suck for organizing knowledge for more than just a dump file (if I want a dump file, I have the most powerful tool of all: Notepad). So, more to come on this front another time. BTW, though Drupal would be better if it had more wikiness, it comes the closest to being the best CMS I’ve used yet. )

Slashdot discussion from Nov, 2004:

Others to examine:
Web Sites, another darling of the blogger set
Simpy which I know nothing about, but looks good and has a Yahoo Group for it: The author is a Java whiz ( so it may become one to watch!
MyBookmarks Dunno about quality
MyHQ Dunno about quality
Spurl Dunno about quality
IKeepBookmarks Dunno about quality
linkaGoGo (because someone linked to this site from there! Referrers do matter! and they are Boston based, so a local company to me)
Sync2it (a list of sites from 1997, perhaps out of date?)
Booby, GPL, which gets the award for best tagline: A Multithingy Something. Like many groupware/cms systems, it stores bookmarks and more but by being lowest common denominator, these all seem to suffer compared to the dedicated tools.
Active PHP Bookmarks once dead now back again?

Onfolio = $30 client side software
Mozilla Bookmark Synchronizer
VisualMarks (formerly commercial, now free, clientside, takes screen shots)
AimAtSite, Seruku Toolbars, Recall Toolbar
Surfsaver (based on AskSam)
ContentSaver Pro (
BKM open source client side bookmark editor, in VB and Python

A nod to the late but not well known Speedle. Made by the folks who invented collaborative filtering, Firefly, after they moved on from Microsoft.$400.

Finally, note that Simpy, like so many others, gets it wrong about the navigation GMail has the same problem. Once you scale, the inability to focus down on a specific subject area in a visual fashion and understand how categories nest becomes problematic. Flat navs are inherently broken, and I can only hope that folks realize this sooner rather than later. We see this now on the new search-engine companies selling combined “index and nav” systems, which allow both structured searching (“Wine: Red: Calif: 2002: etc.”) as well as “traditional” keyword searching, ala EasyAsk. See also my later post, I Hate Tags

Comments? [3]

* * *


On a previous episode...

powered by Textpattern 4.0.4 (r1956)