Eric Lippert
banner
ericlippert.com
Eric Lippert
@ericlippert.com
1K followers 270 following 140 posts
Eric Lippert designs programming languages; prior work includes architecting the Hack and Bean Machine compilers at Facebook, and developing the Visual Basic, VBScript, JScript and C# compilers at Microsoft. Blog at https://ericlippert.com
Posts Media Videos Starter Packs
Doug mocked that up but the title was mine!
There are no senior employees if there is no pipeline for training junior employees, and LLMs will never replace senior employees because LLMs *do not think and will never think*. It's incredibly foolish to destroy that pipeline.
Of course companies are not altruistic. Microsoft provided me vocational training not for the 12 months of terrible quality work they got out of me, but rather to incentivize me to become a highly loyal high quality entry level employee who could then make them 4x salary in revenue. It worked.
The point of giving tasks to interns isn't to get the task done, it's to train the intern! That's why we don't give critical tasks to interns.
Many times at MSFT decades ago I said that QA's job was not testing, it was advocating on behalf of the user. Testing was a tool, meeting user needs was the goal. Who advocates for users now? Seems like no one.
Wait til they find out about XBOX!
The most stunning thing we saw diving Sund Rock yesterday though was the biggest GPO I've ever seen in the wild. It was fast asleep, out in the open, in only 18 feet of water. I've never seen one asleep out of its den and so shallow before.
I spent the morning diving in the Hood Canal and got a pretty decent shot of this beautiful golden dirona. Biggest one I've ever seen, at least as big as my palm.
Since I've started using VS again, ignoring the bad suggestions being constantly made has unfortunately become second nature.

When I was in devdiv my team worked hard to make systems that did NOT suggest that I introduce subtle bugs a dozen times an hour.

Please do better.
It was! And having seen many of my friends and colleagues laid off in the last three years by billionaires seeking to put downward pressure on wages, it continues to be infuriating.
This was particularly galling for me as my aforementioned second rewrite was cancelled after Neal and I put a year's worth of work into, and it was passing almost all conformance tests. The third rewrite blew way, way past the initial timeline. Astonishing to me that Roslyn was not cancelled!
After I was promoted to principal engineer and then given an unacceptable performance review immediately after, one reason they gave was my poor ability to estimate how long a complex task would take. A skill that I was never taught, and was provided no expert mentors to learn from.
When we pitched to VS management that we entirely rewrite the entire compiler and IDE stack for VB and C# from scratch, understandably they wanted to know when it would be done and how much it would cost, so we negotiated schedule and quality goals.
We threw two away when building Roslyn and shipped years late as a result. Though it was ultimately a success that I'm proud of, I still feel that we would have shipped on time had we not thrown away the second one, but rather iterated upon it until it met perf goals.
The first edition of Mythical Man Month suggests "plan to throw one away", but the natural critique is that if you plan to throw one away, you'll probably throw two away. In the second edition Brooks admitted that the original advice was too simplistic.
And then it heard me and shot me quite a look.
Here's the back end of a great horned owl that I shot in Bainbridge Island WA yesterday evening.
I got a shot of the great horned owl that lives at my friend Amber's place. It was hooting at the owl that lives next door when we interrupted it, which probably accounts for the withering look. 🪶
That's a good point. WATCOM invested heavily in optimization but the productivity end was weak. I loved their version of vi though. Used it for years after I left.
What I found particularly odd, having been an intern at WATCOM in both high school and university before going to MSFT, is that WATCOM also had a much better compiler than MSVC, but MSFT thought of WATCOM as an ally and Borland as a threat. I never understood the discrepancy.
There were white papers released with "Buck Forland" on the list of authors. MSFT started heavily recruiting to draw people away from Borland, and got Paul Gross and Anders to join in 95 or 96. It was a weird time.