HN user

siphor

276 karma
Posts1
Comments83
View on HN

Looks like about 20% of comp sci grads in the US are women. Seems like the very simple "why". As for that why... idk probably need to change how kids grow up en-masse. My guess, which probably isn't worth much, is it'll happen slowly, but will probably take a generation or two.

There's probably also some genetically encoded differences on risk-averseness (another guess). FAANG is a great gig and is much more predictable and stable.

I'm not talking about the "trusted by" section, but now that I'm visiting your site again, I see you've added that. When i looked a year back you only had the second section. "The power of open source" which then shows a ton of logos, including: affirm, box, axon, youtube, github, square, shopify, new relic, yelp, weave... If these are all your customers, that really wasn't clear to me! And that's awesome! You should be more explicit that they're all your customers.

I thought it was just saying, "hey, all these folks are using this open source thing that we're also using/selling".

I still can’t believe how much money VCs have pumped into this crap, legitimizing it.

Sequoia and a16z effectively force LPs to put money in it.

Even with a tinfoil hat I can’t figure out their endgame, is it really just to make money off of greater fools?

eh, I understand the sentiment - but it's a little overkill. Junior programmers can be code reviewed and taught these things over time, and also be wildly productive in a production setting.

i get a large majority of software patents should not exist, but are there no software ideas that are truly patentable?

what makes a "physical" patent you consider valid different from a software patent?

I know there is some difference, it's hard to formalize though.

Say someone spent years building a compression algorithm 10x better than everyone else's in some novel way and then tried to capitalize on it. And then google just copied the idea and put this guys company out of business. Isn't that bad?

Hah crazy, it is a scott test. this is the exact packet that he's using. You can see it on the youtube video someone posted below. They also say it turns pink and that's the positive test. But the directions on that site say it has to be blue.. and some nonsense about a bunch of vestibules that the guy just all shakes together it seems? Unclear.

https://www.copsplus.com/sirchie-narkii-test-07-scott-reagen...

Here's the directions on the website:

Confirm suspicions of cocaine presence using the NARKII Test 07-Scott Reagent from Sirchie®. The improved presumptive test utilizes a three-vestibule system to identify powdered cocaine, as well as cocaine bases, crack and freebase. To administer the test, place a small sample of the suspected substance in each of the three ampoules. The first vestibule will melt, creating a blue solution if the sample is pure powdered cocaine. A cocaine base will form hard blue specs that float in a pink solution. Both forms of the substance will dissolve into a pink liquid in the second container. The third ampule will have a burst of blue, that later separates into a pink over blue solution when mixed with either powdered or cocaine bases.

This is hilariously sad. These cops have 0 training, or are out for blood. I'm thinking Hanlon's Razor.

EDIT: Here's the video timestamp where you can see the packet: https://youtu.be/c7j-Ijo2TYw?t=545

Side note HD bodycams are pretty good, surprised you can see and read this packet that clearly.

Might fit under fraud. I hope it does. When I first learned of this tactic, I felt tricked. I assumed with a tip, that I was boosting someone’s salary for a job well done. I’m sure it’s debatable, and that my assumption was wrong? But I imagine most people feel this what a tip is.

Fraud n. the intentional use of deceit, a trick or some dishonest means to deprive another of his/her/its money, property or a legal right.

Interesting on the relevancy stand point. As an iOS developer, I had to learn a whole new set of skills for the interviews... I can count on one hand the number of times I needed to use techniques needed for the algorithm questions. I feel like the skills around making large scale apps is more about being able to reason about large scale projects and keeping everything clean and tidy.

I haven’t really met anyone yet who could talk the talk but couldn’t write code... I’ve definitely met people who have tried... maybe I have a good BS detector, or maybe I’ve been lucky? All people I’ve hired without doing the BST/linked list/DP problems, Ended up working out great. They probably can’t do those problems still, but they can code a damn good app. Guess it goes back to not caring about false negatives

Leadership ability, communication ability, team-work ability are probably more important factors than algorithmic ability in getting a project done, depending on the team. It seems the big-cos have figured that stuff out enough in order to make the development process as much of a meat-market as possible. However, these interviews would probably not select a maximal team for a startup.

The big-cos are also leaving a lot of talent on the table -- which makes it silly when they talk about a "talent-shortage." But, they're also paying enough where people WILL game these interviews in order to get the job... So the best argument I've seen for why these interviews work is that it selects people who work hard at learning this type of problem solving, to get the job... which correlates well with how hard they will work at the job.

Having an "objective" process is near impossible... and in fact probably not something one should strive for. Everything we do has humans has intuition behind it, whether good or bad. I think intuition is very important for finding a maximal team, especially a small one.

"Inclusivity" is kind of weird here.. most often you'll want someone with domain knowledge to get a project going quickly.. It's definitely less important for big-cos, but you still end up interviewing for a web job, an iOS job, etc.. so domain knoweledge does come into play.. at least at the big companies I've interviewed for (google, fb, lyft, uber, etc..)

I doubt this process is risk averse... you'll find people smart enough, but you learn little about their character.

Relevancy seems like a joke in your list... unless all you need is people to iterate on a small set of already solved algorithms in slightly different ways. I'd say that applies to zero companies. Except maybe leetcode, which helps people study for these tests.

The process scales well, and is easy to do, i think thats its strong suit... it's mostly effective across a ton of people.

So in summary, I agree that it seems like the best known process for these big companies. but I doubt it's as risk-averse as you think, and if you truly have a talent shortage, it's an artificial shortage created by your interview process. You might need to take some risks on people if you need to grow faster...

That being said, having that big-co process is great for startups! There's a lot of exceptional talent out there that will fail at the big-cos interviews, but will be better than their engineers for your team :) (There's also exceptional talent that will succeed at those interviews too, but would rather work at a startup) It might be a matter of using your intuition to find them... it's definitely a hard game.

I think this recent data-driven trend that is spreading across all realms of companies is poisonous in some areas where it does not belong... Human intuition still wins in a lot of areas, if it's good.

I would doubt genetic reasons until he tries without all the modern light/darkness. It’s insane how quickly my sleep schedule changes when I sleep in a tent and the sun is unavoidable at literally sunrise... I match the sun schedule pretty quickly, and then get out of it really quickly (1-3 days).

I used to think that but now completely disagree. I’ve been making iPhone apps for too long, and storyboard/ib is so much faster and easier then any of the other options. I only use it for building constraints, and sometimes need to make them programmatically.

There’s a lot of stuff in there though that I ignore because of the issues you mentioned above. If you just keep it to simple constraints it’s great