• (cs)

    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...

  • Minus (unregistered) in reply to Sunstorm
    Sunstorm:
    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...

    This is, simultaneously, the best and worst thing I've ever heard.

  • asdf (unregistered) in reply to Sunstorm

    facepalm

  • MS Ninja (unregistered) in reply to Sunstorm
    Sunstorm:
    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...
    ...then I discovered JRuler.

    But seriously, stopwatches???

  • Max (unregistered)

    They should be happy! The stopwatch plan barely beat out the "one alligator, two alligator, three alligator" plan.

  • (cs) in reply to Max
    Max:
    They should be happy! The stopwatch plan barely beat out the "one alligator, two alligator, three alligator" plan.

    Only after the one mississippi two mississippi plan was scrapped because people felt discriminated against if they were from Mississippi. Thankfully there were no Floridians on the team or the alligator count would have not even been in contention.

  • anon (unregistered)

    Without understanding the purpose of the test, it's hard to criticize the methodology. Maybe, what they wanted to test really was the amount of time a user typically took to execute a save to a floppy disk?

    And who told Mr. Bigshot Tester that, just because he was out of the military, he could start thinking, and having opinions, anyway? I find the charm of ex-servicemembers to be that they know their places.

  • snoofle (unregistered) in reply to MS Ninja
    MS Ninja:
    Sunstorm:
    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...
    ...then I discovered JRuler.

    But seriously, stopwatches???

    Stopwatches make perfect sense. I started my career working with military subcontractors. The right tool was irrelevant; what was important was what tool was handy. After that, once someone made a decision (the intelligence thereof was irrelevant), you had to go with it.

    A hardware engineer I was working with needed a ribbon cable for something, but couldn't find one, so he took a bunch of extension cords, cut pieces of identical lengths, stripped the tips and carefully soldered them to the m/f plugs to make his own 'cable'. Interestingly, it worked; it just looked scary.

  • KG2V (unregistered) in reply to Sunstorm

    sometimes, the manual method works, because your NOT looking for the difference between 10 seconds and 10.1 seconds, or even 11 seconds. Sometimes your looking at 10 seconds vs 1 minute or 2 minutes, and a stopwatch and a human are accurate enough, and there is a stopwatch on your watch. Your just getting some numbers to justify the faster system

  • (cs) in reply to snoofle
    snoofle:
    A hardware engineer I was working with needed a ribbon cable for something, but couldn't find one, so he took a bunch of extension cords, cut pieces of identical lengths, stripped the tips and carefully soldered them to the m/f plugs to make his own 'cable'. Interestingly, it worked; it just looked scary.
    In most cases, I bet it would be faster to go to the store and buy one.

    That, or loot one from a dead or unused desktop somewhere. Were there seriously no corpses laying around?

  • (cs)

    The advantage of the stopwatch approach is that the stopwatch is external to the tested system. Otherwise, you have perfectly precise figures from the tested system itself - so what if the computer has a bug and reports a wrong time?

  • Shinobu (unregistered) in reply to Sunstorm
    Sunstorm:
    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...
    Actually, I do that all the time, because unless the amount of pixels you want to cound becomes very large, it's siginificantly faster than all software based approaches to this problem. Of course it helps that up close the pixels on my LCD monitor are nice distinct squares.
  • John Doe (unregistered) in reply to ammoQ
    ammoQ:
    The advantage of the stopwatch approach is that the stopwatch is external to the tested system. Otherwise, you have perfectly precise figures from the tested system itself - so what if the computer has a bug and reports a wrong time?
    Of course you need to check if the computer stopwatch is accurate with real stopwatches first.
  • (cs)

    Many years ago, I once spent hours trying to match the background of an image with a brush colour in Photoshop. I wanted to put the image on a website as a title at the top and because I didn't know how to crop around the letters, had a rectangle that included the letters and the background from the CD slip cover I'd scanned it from.

    Hours, adjusting the little sliders to try to get that shade of blue that would make it look like the background of the image and the background of the page were the same.

    Then my friend told me about the eye dropper tool...

    -- Seejay

  • The alimentary canal (unregistered)
    Were there seriously no corpses laying around?

    I hate to be the one telling you this, but... intestines does not do a terribly good job at moving electrons.

  • (cs)

    I had to do this about 6 years ago to test 'UI responsiveness' on a thick-client windows app deployed remotely using Citrix. Of course, I also had extensive network traces running to provide actual request-response timings between the clients and the Citrix server (and between the Citrix server and the 'app server'), but management wanted to know how much time between completing the network transactions and the UI refresh. I think I ended up throwing out the stopwatch results and adding some constant amount to the network trace timings to correct for 'rendering time'.

    It was all predicated by a pretty severe WTF: Don't develop and test client-server applications on a 100Mbit LAN and then deploy them to sites with a 56k frame relay without doing a load test or two.

  • Peter (unregistered) in reply to Shinobu

    You can make a quick magnifying glass with a bit of water. Just flick a drop onto the screen, and POOF! instant magnification.

    I used this technique to solve a bet with my brother over how the phosphors were arranged on our TV set.

  • AC (unregistered)
    No, it was none of those things. It was human error. Human error on several levels. At a low level, finger dexterity, at a high level, bad management. To elaborate, R. B. was given special equipment to time these operations; a stopwatch. A MITRE senior engineer put together a test plan involving stopwatches and floppy drive I/O for computers with different RAM and CPU configurations, and somehow got it approved and okayed by her superiors.

    Oops - seems someone left the company in there! I guess that's why you shouldn't have a soccer company doing tests.

    Wait a second... that mitre is all lowercase. So you must mean this federal research lab.

    Ah, I always love to know what my tax dollars are being wasted on!

  • (cs) in reply to anon
    anon:
    Without understanding the purpose of the test, it's hard to criticize the methodology. Maybe, what they wanted to test really was the amount of time a user typically took to execute a save to a floppy disk?
    The fact that the story was submitted here suggests otherwise.

    If you're not being sarcastic, you're really reaching here.

  • BS (those are my real initials!) (unregistered) in reply to AC

    I once applied to work for MITRE. They invited me onsite, flew me in, set me up in an excellent hotel, and gave me a tour of the place.

    I never heard from them again. They spent over $1000 on my plane tickets (I printed a receipt at the gate), but they couldn't spend 39 cents on a rejection formletter.

    Never mind that Travelocity/Expedia/Orbitz/etc listed identical itineraries at a quarter the cost.

    Wasted tax dollars, indeed.

    (My captcha is "bathe" and I find that insulting! :) )

  • dxk3355 (unregistered)

    I just had a job interview with them too. If I interview again maybe I'll ask them about it.

  • Barf (unregistered) in reply to KG2V
    KG2V:
    sometimes, the manual method works, because your NOT looking for the difference between 10 seconds and 10.1 seconds, or even 11 seconds. Sometimes your looking at 10 seconds vs 1 minute or 2 minutes, and a stopwatch and a human are accurate enough, and there is a stopwatch on your watch. Your just getting some numbers to justify the faster system

    And the ignorant method is perfectly fine you're not looking for difference between possessive "you"s and conjuncted "you are"s. You're just getting some words that are phonetically the same...

  • anon (unregistered) in reply to bobday
    bobday:
    anon:
    Without understanding the purpose of the test, it's hard to criticize the methodology. Maybe, what they wanted to test really was the amount of time a user typically took to execute a save to a floppy disk?
    The fact that the story was submitted here suggests otherwise.

    If you're not being sarcastic, you're really reaching here.

    The fact that the story was submitted here means nothing. Private Pyle won't have been the first one to submit a process here, without realizing the process made perfect sense, if you understood the purpose. And, I'm not reaching. Measurements like I describe have been a part of process management since the days of Fredrick Taylor. What would you do- calculate how long a user takes to save something to disk, by using the drive speed? That would be a WTF.

  • (cs) in reply to Barf
    Barf:
    KG2V:
    sometimes, the manual method works, because your NOT looking for the difference between 10 seconds and 10.1 seconds, or even 11 seconds. Sometimes your looking at 10 seconds vs 1 minute or 2 minutes, and a stopwatch and a human are accurate enough, and there is a stopwatch on your watch. Your just getting some numbers to justify the faster system

    And the ignorant method is perfectly fine you're not looking for difference between possessive "you"s and conjuncted "you are"s. You're just getting some words that are phonetically the same...

    Not to mention the plural "youse", as in "hey youse guys!"

  • (cs)

    What's really WTF here is that the test is pointless. Floppy drives spin at standard rates. Even if they didnt there's rarely a spec that the company keep using a particular brand or model of drive. So floppy I/O times are pretty standard on the one hand and unpredictable on the other.

    More useful tests would involve measuring data i/o error rates, eject button longevity, amount of dust blown by the fans through the floppy drive, drive temperatures... but those would require thinking.

  • David (unregistered)

    Any time spent using a floppy disk is too much time. Personally, I think it's been at least a few years for me.

  • Jack (unregistered)

    Sounds like some people finding excuses to get overpaid for crap. Free government money! yay.

  • G Money (unregistered) in reply to KattMan
    KattMan:
    Max:
    They should be happy! The stopwatch plan barely beat out the "one alligator, two alligator, three alligator" plan.

    Only after the one mississippi two mississippi plan was scrapped because people felt discriminated against if they were from Mississippi. Thankfully there were no Floridians on the team or the alligator count would have not even been in contention.

    The alligator and Mississippi systems are so archaic. Everyone knows that there are much more modern systems available, such as one steamboat two steamboat.

  • Mr. Rhee (unregistered) in reply to The alimentary canal
    The alimentary canal:
    Were there seriously no corpses laying around?

    I hate to be the one telling you this, but... intestines does not do a terribly good job at moving electrons.

    How exactly did you test this?

  • IHaveNoName:-( (unregistered) in reply to KG2V
    KG2V:
    sometimes, the manual method works, because your NOT looking for the difference between 10 seconds and 10.1 seconds, or even 11 seconds. Sometimes your looking at 10 seconds vs 1 minute or 2 minutes, and a stopwatch and a human are accurate enough, and there is a stopwatch on your watch. Your just getting some numbers to justify the faster system
    Still it would be much easier and faster (=cheaper) to use a benchmarking software to measure the i/o speed.
  • snoofle (unregistered) in reply to rbowes
    rbowes:
    snoofle:
    A hardware engineer I was working with needed a ribbon cable for something, but couldn't find one, so he took a bunch of extension cords, cut pieces of identical lengths, stripped the tips and carefully soldered them to the m/f plugs to make his own 'cable'. Interestingly, it worked; it just looked scary.
    In most cases, I bet it would be faster to go to the store and buy one.

    That, or loot one from a dead or unused desktop somewhere. Were there seriously no corpses laying around?

    This was in 1986 - not too many spare desktops, and it was a (relatively) secure facility - most doors were locked, so scavenging was difficult. There was no petty cash reimbursement - all had to be done using officially supplied parts (you could get seriously reprimanded for bringing in your own equipment). Besides, nobody cared about expediency - so long as you followed the rules...

  • Doug (unregistered)

    To the person above measuring individual pixels by squinting. If you're on Windows, I recommend installing ZoomIT http://www.microsoft.com/technet/sysinternals/utilities/zoomit.mspx and then just zooming in rather than squinting. If you use a Mac, just hold control and use your mouse wheel. It's built-in.

    As to this whole bit of timing with stopwatches, it's a perfectly valid timing device and test, as long as you're aware of the error introduced. For example, if your tests are on the order of 10 seconds each, and your stopwatch error is as much as a second, you should set the test to repeat 10 times, then perform your timing on the batch. This way, you reduce your error to about a tenth of a second per test.

  • Sharkie (unregistered) in reply to Sunstorm

    The submitting author broke the first rule of the military- which was to question an order or suggest a secondary plan of action.

    As already mentioned, the field soldiers weren't truly made aware of the real function of the test. The test had no real intention of measuring floppy drive performance- this was the cover story.

    The real benchmark was to test human reaction time and the fastest individuals would be secretly offered a position in a new secret squad of stealth ops. where fast reaction time was a strong factor of eligibility.

  • Anon (unregistered) in reply to Ancient_Hacker

    Good thing you added that "Even if they didn't" part.

    The original compact Macs (128K thru Plus) (and yes, Commodore started it) shipped with a floppy drive that took 400K (later 800K) disks that operated in GCR mode. Later Macs shipped with the SuperDrive, which could read/write the 800K disks in GCR mode and the 1.44M disks in MFM mode.

    This is explained in a technote warning users not to try to punch a hole in an 800K DS/DD disk to try to use it as a 1.44M HD disk. People would try this if they had come from the world of 5-1/4" disks (like the Apple ][ before the GS) where you could punch a hole in a SS disk and turn it over to use it as DS.

    When the SuperDrive writes 400K/800K disks in GCR mode, the motor speed is variable, and the disk surface is divided into five zones to allow a constant recording density as the head moves from the outer edge to the center. When using high-density media, data is written in MFM mode, and the drive speed is constant for each track.

    I had two of the GCR drives; you could hear the pitch change as the motor speed varied. ISTR that some guy had written a program that used the different pitches of the drive whirring to make music. (google google) Ah, here's a mention of that technique.

  • Jimmy (unregistered) in reply to Shinobu
    Shinobu:
    Sunstorm:
    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...
    Actually, I do that all the time, because unless the amount of pixels you want to cound becomes very large, it's siginificantly faster than all software based approaches to this problem. Of course it helps that up close the pixels on my LCD monitor are nice distinct squares.

    a screen resulutio of 15x14 works great for this, altrhough I can't see what te hell i'm typing

  • Harrow (unregistered)

    Well, I'm testing third-party software launch times on Vista, so of course I'm not using a stopwatch.

    I'm using a calendar.

    -Harrow.

  • mrs_helm (unregistered)

    Video + timestamps.

    Problem solved.

    It's independent of human slowness. It's independent of the CPU.

  • Grant D. Noir (unregistered) in reply to KG2V
    KG2V:
    sometimes, the manual method works, because your NOT looking for the difference between 10 seconds and 10.1 seconds, or even 11 seconds. Sometimes your looking at 10 seconds vs 1 minute or 2 minutes, and a stopwatch and a human are accurate enough, and there is a stopwatch on your watch. Your just getting some numbers to justify the faster system

    Thus the need to find the average over 10 samples. No?

  • Matthew (unregistered) in reply to IHaveNoName:-(
    IHaveNoName:-(:
    Still it would be much easier and faster (=cheaper) to use a benchmarking software to measure the i/o speed.

    I doubt that is going to tell you what you need to know. There are other things that affect the overall time required to save a file from an application. Filesystem overhead, for one thing. This would involve head seeks (which can be pretty slow for floppy drives) as well as extra read/writes. I say the stopwatch method probably wasn't too bad. Perhaps the reason for doing the timings in the first place was a bit of a WTF, but I don't see anything necessarily wrong with using a stopwatch to measure the total time.

  • snoofle (unregistered) in reply to Doug
    Doug:
    To the person above measuring individual pixels by squinting. If you're on Windows, I recommend installing ZoomIT http://www.microsoft.com/technet/sysinternals/utilities/zoomit.mspx and then just zooming in rather than squinting. If you use a Mac, just hold control and use your mouse wheel. It's built-in.

    As to this whole bit of timing with stopwatches, it's a perfectly valid timing device and test, as long as you're aware of the error introduced. For example, if your tests are on the order of 10 seconds each, and your stopwatch error is as much as a second, you should set the test to repeat 10 times, then perform your timing on the batch. This way, you reduce your error to about a tenth of a second per test.

    On Windows (at least in XP-sp2): start | Programs |Accessories | Accessibility | Magnifier, then set magnification level to 9 - really easy to count pixels w/o squinting!

  • Andrew (unregistered) in reply to John Doe
    John Doe:
    ammoQ:
    The advantage of the stopwatch approach is that the stopwatch is external to the tested system. Otherwise, you have perfectly precise figures from the tested system itself - so what if the computer has a bug and reports a wrong time?
    Of course you need to check if the computer stopwatch is accurate with real stopwatches first.

    What if the computers too fast? You'll get relativistic time dilation.

  • Doug (unregistered) in reply to snoofle
    snoofle:
    On Windows (at least in XP-sp2): start | Programs |Accessories | Accessibility | Magnifier, then set magnification level to 9 - really easy to count pixels w/o squinting!

    Oh, I'm quite aware of magnifier, but the person above mentioned ease of software use, and with Zoomit installed, you hit Ctrl-F1, and you use the mouse to specify zoom. It is much quicker to vary magnification as needed. In addition, it gives you the added bonus of being able to draw on the screen for when you're working with someone else and need to show them something...

  • bisyoinom (unregistered)

    hmm..i remember the time when i tested response times of applications using an online stopwatch :P

    captcha: alarm...clock?

  • einstein-wanna-be (unregistered) in reply to Andrew
    Andrew:
    John Doe:
    ammoQ:
    The advantage of the stopwatch approach is that the stopwatch is external to the tested system. Otherwise, you have perfectly precise figures from the tested system itself - so what if the computer has a bug and reports a wrong time?
    Of course you need to check if the computer stopwatch is accurate with real stopwatches first.

    What if the computers too fast? You'll get relativistic time dilation.

    Either that or you'll just finish before you start - which is great if you like to call it a day and go home first thing in the morning after your coffee :)

  • bisyoinom (unregistered) in reply to Max
    Max:
    They should be happy! The stopwatch plan barely beat out the "one alligator, two alligator, three alligator" plan.

    ...which beat the "one one thousand, two one thousand, three one thousand" plan.

    captcha: muhahaha...took the words right out of my mouth ^_^

  • (cs)
    MITRE:
    R RRIIISSSKKK MMMAAANNNAAAGGGEEEMMMEEENNNTTT PPPLLLAAANNN F FFOOORRR TTTHHHEEE H HH---666000 AAAIIIRRRBBBOOORRRNNNEEE

    Their PDF search engine must have been designed by the same guy.

  • Bob Crankypants (unregistered) in reply to Peter
    Peter:
    You can make a quick magnifying glass with a bit of water. Just flick a drop onto the screen, and POOF! instant magnification.

    I used this technique to solve a bet with my brother over how the phosphors were arranged on our TV set.

    OMG... I can imagine Thanksgiving at YOUR house.

    "Peter, stop staring at the digital clock, grab your brother and come eat!"

    "Can't ma, we're trying to determine the luminance refraction of a digital clock when viewed through water droplets now."

    What a couple of dorks.

  • Northerner (unregistered) in reply to G Money

    The flaw in the plan is that I don't think you should assume anyone from the South to be able to count past 3, much less be able to put words in between the numbers.

    (sick of dealing with southern call centers.)

  • Long Time Lurker - First Time Poster (unregistered) in reply to Spacecoyote

    After reading this one, all I have to say is...

    burp

  • (cs) in reply to Sunstorm

    This reminds me of the time when I used to put my face against the monitor and squint really hard to count the number of pixels of something drawn on the screen. Then I discovered printscreen...

    ... And this reminds me of my fellow colleaque, who had a dead pixel in his laptops LCD screen. He wanted to complain, so he hit printscreen, verified that the dead pixel is indeed visible in the screenshot, and sent it via EMail to the laptop manufacturer.

Leave a comment on “Sir, Seriously, Sir?”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article