Grant Stavely: Blog

1 year, 10 months ago

In 2001 I was answering phones for a company that used OS2/Warp on the desktop, while I tried to find a technical job as the dot-coms crashed.

Warp was six years old at the time, and pretty terrible compared to then-modern alternatives. It was a dinosaur.

Legacy applications and the enterprise directory services infrastructure made the switching costs so high that it would be another year or so before they were able to switch to Windows XP.

Luckily the OS is becoming irrelevant. Ignoring the cloud ~~computing~~ marketing noise: cheap open-protocol traffic tunneling, multi-platform presentation layer client/server authentication and application projection, virtualization, and SaaS offerings are killing the OS market stranglehold. With this lower barrier to entry, expect fun innovations on the desktop. What will change with multi-touch projection walls or tables, universal wireless network access, e-paper, and irrelevant local storage and battery life concerns?

Actually, I’m just looking forward to not using Windows XP. It is a dinosaur.

1 year, 10 months ago

None of us have a truly paperless office. The entire idea is an unreal sales pitch. We do have an mostly paperless office.

Don’t think so? Look at this governmental office scene from the beginning of the 20th century. Those books are not decorative testimonials of owner style and taste to be discussed, or lent to visiting information workers, or even more rarely, referenced. Those filing cabinets are not full of contractual records preserved for compliance reasons. Well, they might have some of those, but mostly they are the operational raw data of the business, and in this place, of the government.

They are the file servers.

More than half the people in that photo are now, in fact, infrastructure. Reassemble this operation with modern equipment and they will be right-sized.

Look how nice those pieces of infrastructure dress for work! My, how low the cubical walls are, why it’s almost the revolutionary open floor plan style office I’ve read about tech companies using in so many business articles!

Are all the file drawers open as staging for the photo, or is that the actual equivalent of locking an open ~~file folder~~ directory?

Unrelated and cool: Note the ghosts in the farthest depth of the picture.

1 year, 10 months ago

The Harvard Business School Working Knowledge blog has posted an interesting paper about the gambler’s fallacy, the notion that when presented with random binary events, we tend to identify patterns which we believe will impact future events. As an analyst, this is very interesting.

I wonder if after reviewing six false positive events, having only heard about true positives from third parties, can reaction time to future alerts be expected to slow? Where F is a false positive and T is a true positive, is TFFFFFFF to be handled differently than FFFFFF? Is it incorrect to do so? The paper relates instances where it is correct to subscribe the gambler’s fallacy, and surely identifying a systemic form of false positive with no change to alerting controls is an adequate example.

But what about non-systemic repeat false positives? An alert looking for something simple like a secret word, could first fire on an e-mail, then an instant message, then a file transfer, then an inbound connection to a web server, and all from unrelated sources. An analyst would be wrong to assume the gambler’s fallacy only if the secret word is not so secret after all. If it is indeed secret and the alerts were coincidence, what of no similar alerts for a month, followed by another string of false positives?

What if the analyst can reliably expect a true positive and knows it is only a matter of time? I expect that would prevent the gambler’s fallacy from effecting reaction times, as it is nearly equivalent to being presented with FFFFFFFTFTT after the events have transpired.

Unfortunately unlike experiment and true binary random chance, analyst’s both handle events they can reliably expect, and have control or tuning capabilities of alerts. An analyst can and should tune a control’s or alarm’s sensitivity or volume to prevent FFFFFFFF occurrences but always at the risk of squelching FFFFFTFFF detections.

1 year, 10 months ago

The SMTP RFC’s have been replaced! A few things from the list caught my eye:

No spaces are permitted after or before the ‘mail from:’ colon.
587/submission is back in play and recommended
Lots of striked-out garbage
Post DATA 550’s!

The previous RFC on all but the ‘mail from:’ items have been ignored by most sysadmins for security reasons.

is hosted in Amazon's EC2. All of my content, unless otherwise noted, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License (what a mouth full!). The articles are syndicated in Atom, and RSS formats. The microblog and photos are kept in separate feeds.