Misc. Ramblings

Week of 20 March through 24 March 2000
Last Week    Mon    Tues    Wed    Thurs    Fri    Next Week
Home     Diary Index     Search
Email Dan    
Jump to Last Update: Friday 7:00 am HST

Monday - Spring has Sprung - 20 March 2000

Blowing Smoke. So, you know all that money ($246 billion USD) that the smoking industry was supposed to pay? You know, the money that the state legislators have been salivating over and spending before its been completely distributed yet? Well, what is being seen as a possible strategy to reneg on the deal, the tobacco industry is rumored to be working on filing for bankruptcy protection.

This, in the face of possible damage awards from a class-action case filed in Florida. Now, you have to understand that this could just be a way of trying to influence the Florida case. And of course, the industry is not commenting. But the states are taking this so seriously that they will be looking at retaining their own bankruptcy attorneys just in case.

Dual Personalities. Who said you couldn't be in two places at one time? Researchers at the US National Institute of Standards and Technology published their latest findings in the January 20, issue of the British journal Nature. In it, they confirm that "a single atom can be simultaneously located in two separated places (a physical state termed a quantum-mechanical superposition)." However, in the real world, you don't see such phenomena because as the distance increases between to two locations "the influence of the environment on the delicate superposition—quickly become a factor as you increase the distance between the two states of the atom. This dependence on separation tells us why it’s impossible for us to see superpositions in our macro-sized world,' explains David J. Wineland, a member of the NIST team." See the NIST story here.

***** Noon Update *****

Reading is Good. That's right. You heard it here first. OK, maybe not. But reading is good for you. So check out the Daynoters. I do. Every weekday, and sometimes the weekend too! Topics span the globe from Christianity to computer hackers.

Fir' 'zample. Doc Crider and Dave Farquhar have a few words about their faith. But if that offends you, check out Dr. Keyboards site. You may see an advertising banner like the one I saw this morning: Eclectic Witch. Online Auction for Pagan/Wiccan [whatever the heck that is] merchandise. "List your items for FREE!" Or, for an interesting story about hackers, Shawn Wallbridge has a link to the University of California at San Diego. UCSD set-up a box stock Red Hat Linux 5.2 server on the net and watched as the script kiddies did their thing.

I haven't listed the entire gang here, but that doesn't mean anything other than I am too lazy to do so. Each and every one of them is interesting and informative. So do what I do, go to Daynotes.com and read all of them. It will be worth your while. Honest. Or your money back. Your mileage may vary. Objects are closer than they appear. The door is ajar. The dog ate my web page.

Aloha!

Top / Home / Monday / Tuesday / Wednesday / Thursday / Friday


Tuesday - 21 March 2000

SpamCop. Dr. Pournelle's column at Byte.com deals in part with spam (unrequested and unwanted email, not the bad for your arteries meat-like tissue sold in the blue and yellow can). And one of the tools he uses against spam is SpamCop (the original version of the column had an incorrect URL. It has now been fixed. If you d/l'd the page earlier, the correct URL is www.spamcop.net). SpamCop seeks to stop spam by determining where the spam came from, and then automatically sending a notice to the appropriate sysadmin requesting the termination of the account. Obviously, this will only slow spammers down a little. But at least it's something that you can do. And if enough of us take the time to send four or five of the worst spam that you get each day, maybe it will make a difference in the long run. If not, at least you'll feel better by doing something, rather than just letting your blood boil every time "You've got mail" means you've got spam. Think about it.

Opera 4.0 Beta 1. For those of you who are adventurous, the first beta of the Opera 4 browser is now available here. I don't feel like testing it so I haven't seen it myself. But feel free to give it a try. For you Linux lovers out there, version 4.0a Preview 2 is also available. As with all beta ware, you must assume that there are bugs which will cause it to blue screen on you. So don't come crying to me when it does.

Speaking of browsers. Mozilla, aka Netscape 5.0, ahem, make that 6.0, is still in alpha. This, after almost two YEARS to the day (March 31st). They are hoping to release their fist beta Real Soon Now. But at this point, it's irrelevant. So what's the point guys? Check them out here. By the way, if you live near Ess Eff, they are having an anniversary party on April 6th (see the flyer here).

***** Noon Update *****

MS Office2000 SR-1. One of the Daynoters has inadvertently run into problems after installing MS Office2000 Service Release 1 (SR-1). If you haven't installed the patch yet, it is highly recommended that you hold off on that until it can be determined what is the source of the problem. On the other hand, if you have already installed SR-1 and have had no problems, I would interested in hearing from you.

Aloha!

Top / Home / Monday / Tuesday / Wednesday / Thursday / Friday


Hump Day Wednesday - 22 March 2000

Thank you to Dan Bowman for the link to Jakob Neilsen's current Alertbox. There, Neilsen postulates that under certain circumstances, a sample size of five is sufficient for the purpose of studying usability problems in web pages. Hmmm. Maybe so. Neilsen, as far as I know, is the expert on such things.

But I note, if you were to ask a research statistician what the minimum sample size was for just about anything you wanted to study in this great wide world, five would not be the number. In fact, under most circumstances, you need several times that to get even the most basic, statistically valid conclusions.

For example, I use a simple web-based calculator, running over at the University of California at Los Angeles (UCLA) when I am designing a study and need to know the minimum number of respondents I need (see it here). The two main variables in using the calculator is what kind of statistic are you looking to use (mean, proportion, or population total), and what is your estimated population size. Be aware that, in general, as you get into smaller population sizes, your sample size soon becomes the same as the population itself (that is, you have to sample everyone in the population).

But for simplicity's sake, let's say your population of users is equal to 100,000. And you wanted to find the average score of some variable in that population (like how easy it is to use your web's navigational page). According to the generally accepted formula dealing with this question, you would need to sample, that is ask, 383 people. Hmmm.

Let's try a smaller population. Like the number of unique URLs that hit my site last month (about 200). So I plug in 200 as the population and ta da! I only need to ask 132 of the possible 200 people (remember, as population size decreases, the required sample size, as a percentage of the population, goes up). Again, I don't doubt that his formula describes what he says it describes. And yes, there are statistical analyses that can work with sample sizes down into the 20 range. But. I'm not so sure you could use the information you get by using his formula and still have a statistically valid result. I caution readers that I may be comparing apples to oranges since I don't know how he arrived at his formula. But I also caution people that on the face of it, five just doesn't seem to be enough to base your e-commerce life on. You decide (and thanks again to Dan Bowman for the link).

True Colors. The local small business association Republicans showed their true colors yesterday. The head of their association was interviewed on local TV regarding the US Supreme Court ruling on the Federal Food and Drug Administration's (FDA) control of tobacco. As you may know, the Court opined that Congress did not intend to have tobacco considered to be a drug and therefore the FDA does not have jurisdiction over it. So, what did the man say (and I'm admittedly paraphrasing here)? He said that now that the restrictions against selling to minors would no longer be enforced (the FDA funded monies for enforcement), his members could now start selling to minors [without fear of being harassed by the police] and make some real money. Yes, he actually said that. On TV. The difference between Republicans and Democrats could not have been made more clearly.

In a later story about the Democratic Governor wanting to privitize, among other things, one of the boat harbors. One Republican State Senator was shown on TV saying he was against it. Now why, of all people, would a Republican be against making guv'mint smaller? Because his rich Republican constituents have their yachts docked there and they know they've been having a free ride having the State run the place. Given that a private company would actually start charging them what it costs to do so (instead of the State subsidized rate), they are up in arms. How dare the Democrats try to take away their welfare, I mean boat benefits! This is unfair [true, not too many poor people have $100,000 yachts sitting in the harbor - Ed.] So the Senator is trying to spin the story to be one of arrogance on the part of the Governor for not listening to the Republicans. My heart bleeds for them.</sarcasm>.

As an aside, it is interesting that the Republican owned morning paper doesn't mention a thing about either interview. I wonder why?

View from the Top. At some point, I've always figured I would go back to school and get a Masters Degree. But the question was always when and what kind? Well, I've decided now would be a good time to try. So, I've put in my application at the University of Hawai'i at Manoa in their Masters of Public Administration program. I've filled out and sent in the applications (there are two of everything), requests for transcripts, letters of recommendations, and various other and sundry forms. So far, the applications office has replied that the application is complete and I meet the minimum qualifications (Bachelor's degree, college GPA of no less than 3.0, blah, blah, blah).

Tomorrow afternoon, I am scheduled to go up to the University to be interviewed. I am not the most gregarious [to say the least - Ed.] of people so this will not be easy for me. Any good thoughts that you can send my way would be appreciated. Thanks in advance.

***** Noon Update *****

Office2000. No. Not that one. In celebration of the release of MS Office SR-1 (or as Dan Bowman called it, Release from Hell), we are ordering six copies of Corel's WordPerfect Office2000. For the foreseeable future, we will keep both MS Office and WP Office running side-by-side. Then, at some point down the road, we will determine which has the features we need while at the same time not causing things to fall over every time it is updated. Who knows, Linux may be next.

Mail Call

----- Original Message -----
From: Dan Bowman [dbowman@americanambulance.net]
To: Dan Seto
Cc: Jan Swijsen [sjon@svenson.com], JHR [culam@micron.net]
Subject: Perhaps apples and kumquats
Date: Wed, 22 Mar 2000 11:27:56 -0800

Apples and kumquats? For statistical sampling, the larger the population, generally the better the information; for usability studies, especially initial testing, maybe not. I say that because most of the big roadblocks to functionality should show up right away: links that don't work; failure of logical flow; forms that will not send; and just plain getting lost in the system. A small, focused group who are not involved in the programming should catch most roadblocks; then, reprogramming and re-testing with a subset of the same group plus a few fresh faces should catch the next round of problems: those that did not show on the original test because users could not get to them and those introduced by the programming changes.

That's how we've been handling it at the work site with our applications and website development. So far; so good. I don't know if we're doing it correctly, but we don't have a large population to work with. (I've copied this to Svenson as he's in development and to JHR as he may share the sample size of his testers.)

Um, and take the period off the end of the link as posted; it breaks it in IE5 (testing sample size=one <g>),

Dan

----- Original Message -----
From: Dan Seto
To: Dan Bowman
Cc: Jan Swijsen; JHR
Subject: Re: Perhaps apples and kumquats Take II
Date: Wed, 22 Mar 2000 09:48:43 -1000

Hey Dan,

It probably is a case of apples and mangoes. But he does note something about population homogeneity. To wit; "You need to test additional users when a website has several highly distinct groups of users. The formula only holds for comparable users who will be using the site in fairly similar ways." In other words, if the site is an intraweb site, and the only users are White, Anglo-Saxon males, age 40 some-thing, then five may work. But population statistics makes no such assumption. Hence, the need for a larger sample size. Notwithstanding that, he may still be right. I don't know. I'm not a research statistician (and I don't play one on TV).

Aloha!

Top / Home / Monday / Tuesday / Wednesday / Thursday / Friday


Thursday - 23 March 2000

Live High, Train Low. Sounds like something from "High Times." But actually, it's the latest in physical training methodology. A local company, called Labman Hawaii, Inc. is doing research of how efficacious it is for elite athletes to live at high altitudes, but train at low elevations. It is postulated that VO2 max, hemoglobin and hematocrit counts should increase. It is further believed that if these measures increase, that athletic performance should also increase.

So, 25 diverse professional athletes from the US, Russia, Australia, Hungary, Canada, Yugoslavia and Czech Republic are living on the volcanic slopes of Mauna Kea at the 6,800 ft. level (2,073m). While at the same time training at sea-level. VO2 max measures an individual's maximal volume of consumable oxygen, or their endurance. Hemoglobin and hematocrit are measures of concentrations of red blood cells, which carry oxygen through-out the body. But then, you knew that.

You also know that, while resting at high altitude, the body compensates for decreased concentrations of environmental oxygen by producing more red blood cells. This in turn enhances oxygen delivery to the muscles. Which, it is believed, should increase their performance.

While there is already an Olympic training facility at altitude in Colorado Springs, they do not have the opportunity to train at sea-level. But, in Hawai'i, you can. And more importantly, you can do it with ease all year-round.

More mail on the sample size debate from yesterday:

----- Original Message -----
To: Dan Seto
From: Jan Swijsen [qjsw@oce.nl]
Reply-To: [sjon@svenson.com]
Subject: Re: Perhaps apples and kumquats Take II
Date: Thu, 23 Mar 2000 09:17:31 +0000

I don't see a definition of usability anywhere in the text. That is a first problem. For getting bugs and errors (ex non functional links) out you don't need many users. But, imo, bugs are not a usability issue. Colours, fonts, attributes, layouts and program sequence make up usability issues. A bug may be bad for the usability of the site but can be solved definitively. A usability problem is often not solvable in a definitive way, satisfying all uses. Links that don't work are not usability problems, links that work but are invisible are usability problem. Having to fill in a credit card number before selecting products to buy is a simple usability issue (ie most people don't accept it and those that do probably have no problem with entering the credit card data after selecting a product). Having graphical buttons is a more difficult usability problem (some users cannot use buttons some users find button-less applications too bland, either way you lose users).

A second and bigger problem is that we are talking about computer usability here. If he says 5 users are enough he is missing a point he states himself later ie "you need additional users if you have highly distinct groups of users". If you know your target intimately, as on an intranet you have only one group of users. Out 'in the wild' however you have got Mac's, Linux boxes, Win9x users, rusty old i486 boxes, Pentium II, III or IV screamers. You must test for small screens and large screens, with notebooks and desktops, there is MSIE and Netscape etc. No test with only five users will cover that type of variability and demanding each of the five users to test in an environment they are not accustomed to will screw your results, ex ask a Mac users to test the usability of a site using Netscape on Linux and most the remarks you get will have nothing to do with the site but more with OS differences.

Another problem (but related to the previous one) which he doesn't touch is the choice of the five testers. He does mention that you could recruit "15 representative customers". I assume he implies that the 5-user-test should be done with 5 representative customers. You are obviously not doing the test right if you select 5 iMac users if most of your clients are running NT. The biggest problem (cost) of most statistical tests is determining what are representative users and what is the statistical significant sample size. It is also the most abused part of most statistical research.

He succeeds in making a case for focused testing as opposed to extensive testing. But 5 users isn't the magical number.

Aloha!

Top / Home / Monday / Tuesday / Wednesday / Thursday / Friday


Aloha Friday - 24 March 2000

It's Friday!

Calling the Odds. The newspapers are rife with rumors (again) of an impending settlement between the US Dept. of Justice (DOJ) and Microsoft (MS). But, I've said it before and I'll say it again. If MS were going to settle the DOJ case it would have done so already. The great majority of cases that are settled, do so before the trial begins. But. Once a trial does start, it is a very rare bird that ends in settlement this late in the game. When it does occur, it is only when there is a "slam dunk" where one side or the other comes up with something so great that you have to settle.

I know some other columnists periodically come out with predictions that MS will settle because they have seen the signs. Read the tea leaves. Thrown the chicken bones. But so far, MS hasn't. And my guess, for what it's worth, is that they won't. Why? Because there isn't any upside for MS to do so. To settle now would be to agree to almost everything the current Judge and DOJ want now anyway. So why bother? Better to take your chances at the appellate level. On the other hand, if the DOJ blinks first, and agrees to not split MS up into mini-Bills, then maybe MS might settle.

Further, it's the Judge and the DOJ who keep crawling to MS crying that "the sky is about to fall, the sky is about to fall! Really! We mean it this time. We're not joking!" If it were the other way around then I might think there is a chance. But it ain't.

I could be wrong, but I don't think so. [postscript - I wrote the above yesterday afternoon. This morning, the paper says the DOJ has retreated from its demand that MS break into multiple entities. Blink it did. Change things this does. If they retreat further, there could be a settlement. But if they do, the next question is how strong a case did they have in the first place?]

The Devil You Say. The flap over MS Office2000 SR-1 is both less and more than at first glance. While one of the Daynoters did have substantial problems with SR-1, some others have not (myself included). While I do not enjoy downloading 27MB "service" updates, I did not have any extraordinary problems doing so. And once installed, I did not have any problems running any of my Office2000 applications.

However, I do not appreciate MS coming up with a new registration wizard in their retail and OEM versions of Office. And, even though I don't think I will have to re-install Office 50 times before the next release comes out, who knows how reliable the auditing system is and how easy it will be to install if you don't have internet or phone access at the time of the install. The devil, as they say, is in the details.

Interview. The interview before the admissions person at the University of Hawai'i went well (I think). Mostly general questions about why I want to go there and what I want to get out of it. I am hopeful that I will be accepted into the Masters of Public Administration program. I will know for sure some time next month.

Aloha!


Last Week    Mon    Tues    Wed    Thurs    Fri    Next Week
Diary Index   Link to the Daynotes Gang

© 2000 Daniel K. Seto. All rights reserved.