I like realistic posts like this. She presents her experience and the problems associated with trying to diversify a tech team. She's sincere in her efforts, yet still finds she can't achieve her goals. Also, she brings up two points I've long believed about the troubling lack of women in tech.
She's 100% correct about the fact that it's too late to make a difference in industry when you're only dealing with women who just graduated from college. You have to start earlier and teach young girls that technical skills are necessary. Even if they don't like technical subject matter, they should be made to learn it. It's no longer an option.
Aiding men in industry who want to help is crucial. I see a lot of rhetoric online about how all men in tech are a problem and that they should be excluded from this whole process. "Men can't speak for women." If you believe that, you're contributing to the problem. Perhaps men can't speak for women, but they can at least speak with them.
Patience will be required because this situation isn't going to be solved overnight, or even in 10 years. It'll take a generation of concerted effort to really make a difference.
I've run into this exact situation. I have a C extension that does some specialized vector operations. These functions use some language models that can be fairly large. The language models are fixed throughout the calculations, and the other
data required can be segmented so the operations can happen in parallel. Also, the language model and data can be handled as read-only.
Trying to do this with the multiprocessing module is not terribly straightforward.
This is what Walmart has been doing for quite a long time. They will essentially take any return and charge the vendor for it.
My mom used to work as a bookkeeper for a garment company and the owners were thrilled when they landed Walmart. She said dealing with Walmart was a full-time job and they were ruthless in their tactics. They started to receive a large number of returns, and Walmart would deduct the cost (item cost, shipping, processing, etc) from their payments owed. I believe she also said they were Net-90. So you basically ship them 3 months worth of goods and you're left holding your dick for a while.
Most of the items they received back weren't in saleable condition, so it was a loss. The company owners decided to let Walmart deal with those items instead of having them shipped back. It turns out Walmart was donating those items and getting a tax write-off. When my mom figured this out, she suggested that her company do this instead of letting Walmart do it. It helped offset some of the costs of doing business with them.
The point is, when you deal with a distributor like Amazon or Walmart, it can bankrupt you if you aren't ready to deal with the volume and draconian policies. So these small businesses are right to be pissed and scared.
I've been working with speech systems on and off since 1998. They're all pretty bad in my opinion. They're not even almost there yet.
I remember studies from years back that speech is far lower throughput than typing for nearly every language except Chinese.
I hate talking, especially to a computer, when I have a typing interface. Speech is such a serial, synchronous information channel. That fact paired with how context-limited speech systems are makes for a very poor experience. It's really going to take an inventive design to get people to use speech regularly for machine interaction. We're just not there yet.
Also, if you hear the term "refactoring" too much and the system being built still does not work, then that's probably a good indicator of bullshit. Engineers (and managers) need to be taught to resist the urge to refactor before the system solves the problem. Once you've solved the problem, you've connected things end-to-end, and only then will you know what refactoring makes sense.
Ugly code that works > pretty code that does not work
I took the article context as a non-technical manager wondering why a project wasn't complete after using a good deal of resources. Usually "bullshitting" refers to the tactics engineers, or employees in general, use to explain why they aren't done when they should be. If engineers are using these phrases and projects are being completed, then you can assume they're not bullshitting.
If you're a competent programmer who uses these terms and phrases, then indignation is not an unreasonable reaction.
However, I think it's a completely valid article if you've ever experienced the interactions between an unproductive programmer and their non-technical manager. I've witnessed this dance every week for the past 3 years. The programmer in this case has used every one of those phrases except "platform independent".
When I ditched cable TV about 7 years ago I lost track of these outlets. I can't say I've missed them. I occasionally open cnn.com if I'm bored, mostly because it's quick to type.
I have a (physical) subscription to the Economist, which is where I get all my serious news. I can't tell you how great it is to read the news without the page content shifting all around because of ads loading.
Having been a "salaried contractor" for my second job, I learned to avoid those positions like the plague. When customers stop buying the company's software, guess who's the first to get the ax? Also, you get to sit by and watch project managers agree to all sorts of outlandish features that you'll be stuck implementing.
I interviewed there in 2012. The engineer called me, did his little talk and we jumped into the questions. His first question was the subset sum problem. The literal problem. It wasn't even couched in a practical application. So I said "this is subset sum. You can solve it with dynamic programming. I would look it up in a book." His response was "but I want you to solve it." I stopped the interview and thanked him for his time.
I'm a senior engineer. Those types of interviews are to stoke the egos of green engineers who remember the solutions to their algo finals questions. If I can look it up in a book, it's not worth the time to quiz me on it.
And I would bet in a large subset of those cases it's because the person implementing those tests and the person interpreting the results aren't the same. And that neither understands basic statistical hypothesis testing.
She's 100% correct about the fact that it's too late to make a difference in industry when you're only dealing with women who just graduated from college. You have to start earlier and teach young girls that technical skills are necessary. Even if they don't like technical subject matter, they should be made to learn it. It's no longer an option.
Aiding men in industry who want to help is crucial. I see a lot of rhetoric online about how all men in tech are a problem and that they should be excluded from this whole process. "Men can't speak for women." If you believe that, you're contributing to the problem. Perhaps men can't speak for women, but they can at least speak with them.
Patience will be required because this situation isn't going to be solved overnight, or even in 10 years. It'll take a generation of concerted effort to really make a difference.