Steven Sinofsky Profile picture
Jan 2, 2022 16 tweets 7 min read Read on X
061. BSoD to Watson: The Reliability Journey …rdcoresoftware.learningbyshipping.com/p/061-bsod-to-…

Super excited to share this set of stories about software quality, I mean blue screens & crashes. Ever wonder what happens when you click "Send to Microsoft?" Does it matter? Where's it from?Who invented it? 1/
2/ In the Windows 95 / internet era when so many people started with computing, "crashing" was a thing computers just did. You'd be working away on a word processor or paint program and 💥 the PC would freeze or worse.

This happened on Macs too. Mac had a very graceful fail :-) Mac Sorry a System Error has Occurred error message.
3/ When I started writing, I wanted to go through the entire history of how the PC handled crashes. But along the way, I realized what was fundamentally a user-hostile event just got more hostile over the years. Unrecoverable application error.Application Error General Protection FaultProgram has performed an illegal operation.A fatal exception has occurred.
4/ Who was General Protection in what army?
How "illegal" was the "operation"?
Was the exception really "fatal"?

These and other Ups led to internet "memes" (not the phrase back then) like this Tools Options Crash.

And of course the defining "Blue Screen of Death" in NT 3.1 Tools Options Crash with options such as "crash every 2Original NT 3.1 blue screen "**STOP" error. Full b
5/ Turns out back during Windows 3.0 a clever engineer on the Windows team invented a tool called "Dr. Watson" (originally called Sherlock, but that name was used though a decade before Mac).

Watson captured some minimal but critical system info that could be shared w/Microsoft. Dr Watson icon
5a/ Closely related, though not obvious, was development of sophisticated "Undo".

A big problem with s/w was how "destructive" operations (editing!) led to defensive use (saving, backups, etc) which stressed the system. Undo was a first step in dramatic quality improvements. toolbar with undo/redo buttons
6/ There was no internet quite yet so it sort of sat there for a few years.

Then right around holiday 1998, an engineer on Office (a hacker's hacker, Kirk Glerum) had the insight to connect Watson to the internet. he wrote a memo, which was weird b/c he mostly wrote MASM. scan of original D.W. Spec about sending a crash to Microsof
7/ Just a brilliant idea. It seems so trivial / obvious now, but before then software didn't do this (there were examples of copiers that signaled errors over phone lines and mainframes did some of this).

Very quickly the team jumped on this idea. Every crash was a data point. Web page image of a crash.
8/ Many more details about how this changed our culture, what was interesting about in terms of customer experience.

I always thought of it as a dramatic change in *computer science*. It turned fixing bugs at massive scale into a solvable problem. My college recruiting prez. Slide "Computer Science Will never be the same"
9/ The most important thing we learned quickly was the 80/20 rule—80% of the crashes happening (in the real world) were caused by just 20% of the bugs. In fact just a few bugs were *half* the crashes. This "Pareto" distribution was dubbed "the Watson curve". Graph of the Watson curve
10/ The team even wrote a paper that was published in Communications of the ACM "Debugging in the (Very) Large: Ten Years of Implementation and Experience". First page of the paper mentioned
11/ This was the start of a dramatic change in software quality (yes I realize people will make jokes!) It is hard to put into words how fundamental this was to software engineering.

Of course this is part of every mobile platform today but it is amazing to think of the start.
12/ I tried to capture many of the details of this evolution. It was pivotal point in the history of engineering brought on by the internet.

We followed this with "watsonizing" everything: feature usage, help topics, spelling, and more. (Prez from 2004)
13/ I wanted to use this first post of 2022 to thank everyone who has been along for the journey of "Hardcore Software". I can't thank enough the over 200,000 unique readers. It is so amazing to be able to share these lessons and history.

Thank you 🙏
14/ Please consider subscribing for 2022. We're only halfway through and will soon be covering topics like SharePoint, .NET, NetDocs (!), the Ribbon, Courier (!), Windows 7, Windows 8, Surface, and so much in-between. hardcoresoftware.substack.com
PS/ I love the personal reflections shared by the members of the teams in "Hardcore Software". Here is Kirk Glerum on what it was like to build and stand up Watson. (ignore the "Mr.", that's Kirk 🙃).

Subscribe and check out more comments. 	 Kirk Glerum 10 hr ago  I spent ten years, nearly half my c

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Steven Sinofsky

Steven Sinofsky Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @stevesi

Mar 13
Today OpenAI put out an AI Action plan for the government.

OpenAI’s proposals for the U.S. AI Action Plan
Recommendations build on OpenAI’s Economic Blueprint to strengthen America’s AI leadership.

It is a broad document that covers many topics. I want to highlight copyright. Image
In the document there is a call for permissive AI training, noting the carveouts that exist in other countries. Summarized is a call for a "Freedom to Learn".

I mean who is against freedom? Against learning? Against a freedom to learn? No one. Except... cdn.openai.com/global-affairs…Image
First notice how much they have pivoted to China away from the "please regulate us to save jobs". This is noteworthy. Still this is a call for regulatory capture. Just a different mechanism. Instead of blocking others it is enabling others on the assumption scale wins. Image
Read 43 tweets
Aug 12, 2024
Early-Childhood Tablet Use and Outbursts of Anger - // This study is out today and getting broad media play piling on to more evidence tech is bad. It is so dumb a read it should be getting coverage for how bad it is at science not as a way to "exercise caution in tablets." 1/Image
Image
2/ This study appears to be instantly generalized to all screens and all content. The underlying tool used to measure the use of *tablets* is a survey+tool described in this referenced study ("CAFE"). Today's paper clearly takes a subset of this data, which is itself questionable.Image
3/ The underlying study as of 2020 says it monitors content on Android devices. I can't tell and there's no open data available to confirm that for today's study (that I could find.) This is a minor point perhaps but has obvious issues with sampling. But there are other issues:
Read 12 tweets
Mar 21, 2024
Today the US DOJ+16 states/DC (HA we had 19+) filed suit against Apple over abuse of market position ̷b̷y̷ ̷m̷a̷k̷i̷n̷g̷ ̷a̷ ̷v̷a̷s̷t̷l̷y̷ ̷b̷e̷t̷t̷e̷r̷ ̷c̷o̷m̷p̷u̷t̷e̷r̷ in an effort to keep customers reliant on iPhone.

🧵contd until I lose steam COMPLAINT In 2010, a top Apple executive emailed Apple's then-CEO about an ad for the new Kindle e-reader. The ad began with a woman who was using her iPhone to buy and read books on the Kindle app. She then switches to an Android smartphone and continues to read her books using the same Kindle app. The executive wrote to Jobs: one "message that can't be missed is that it is easy to switch from iPhone to Android. Not fun to watch." Jobs was clear in his response: Apple would "force" developers to use its payment system to lock in both developers and users on its platform...
1. This is scary/concerning/freaky if you work at Apple. My first thoughts go to them. What I can say is heads down, be patient. It’s an ultramarathon.

2. If you are a competitor cheering then history tells us down the road you will either become a faded memory or will be sued.
Of course I am not a lawyer and don’t pretend to be one. When the Microsoft case first started in the early 90s. MS’s GC said to me “you have to remember, people who chose to practice antitrust (AT) law not only believe in it but see ‘monopolies’ and ‘abuse’ everywhere.”
Read 50 tweets
Sep 25, 2023
So much of the evolution of technology can be summed up by “what’s new, was already done before…but being first (or early) if often no different in result than being wrong.”

Of course being done before is never ever the same as the new things… 1/
2/ New things that appear to have been done before have a different perspective, bring unique market forces to a problem, and rely on technologies that are often more mature, not brand new.

Many latest and greatest inventions fail and need to be reinvented in new contexts.
3/ I lived through too many Microsoft examples where we were “first”—even “innovative”—only to watch other companies come along and capitalize on something conceptually close/identical but implemented entirely differently.

Their patience and choices made all the difference.


Image
Image
Image
Image
Read 4 tweets
Sep 13, 2023
Apple's 'Mother Nature' sketch was a complete dud, and didn't belong… // No, no. Issue is much more subtle and practical. Need to separate weird marketing from reality. This is greenwashing but the green is…profit. This isn't Bud Light. Or even "woke" 1/appleinsider.com/articles/23/09…
2/Sure the presentation might have been awkward or even a dud to some. A quasi-religious tone viz. Mother Nature isn't everyone's approach.

At the same time, every fact or position put forth is a strategic, margin-positive, and innovative effort from Apple. Super important. Mother Nature in video.
3/ Start with packaging. Most people haven't thought much about packaging. Even most who have made something needing a package haven't thought about it. Packaging is *expensive* and necessary. It is also a whole discipline. How many knew you could get a PhD in packaging? About The only packaging Ph.D. degree offered in the United States. #1 in Packaging programs (Source: universities.com) The School of Packaging (SoP) holds the prestigious reputation of being the first, the largest, and the most comprehensive packaging school in higher education.  The School educates students to thrive in careers such as packaging engineering, design, manufacturing, research and development, and sales.  Our graduates can be found across most industries and global Fortune 500 companies, and our Ph.D. graduates teach at universities around the world. Students graduate with fo...
Read 24 tweets
Jun 6, 2023
Why are people so quick to proclaim failure for new products? It seems a dumb thing to ask. I mean knowledgable people look at a new product and think it doesn't cut it and will fail. Much more going on. Innovation is nearly impossible to deliver. Harder to predict/analyze. 1/ Scott Galloway @profgallowa...Making the Laptop Commonpla...Apple should pull the plug ...Tesla Motors faces rough ro...
2/ Regardless of the era, predicting failure has always been easy, always been attention grabbing, and always kind of fun. Some say it's necessary simply to counter the marketing and power of the launch. Silly. A launch still has to battle the market. The market is really brutal.
3/ Predicting failure is a form of social credit, a way of elevating oneself above the company. It is in effect a power grab. It is also a form of grift. A con. These are harsh words but let me explain.
Read 24 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(