Hacker News Reader

Comment by vcistan-inmate | Hacker News Reader

I don't understand how YC can fund non-profits like VotePlz and the ACLU while not funding stuff like this. All of the startups funded by VCs use free software, often exclusively, but these VCs continue to refuse to adequately fund its development. Marc Andreessen even publicly boasted about how much OSS his companies use[1], which he of course doesn't pay for.

I also shudder to think of how Eaton will fare on the job market should he actually be forced to seek regular employment. Will he be whiteboarded? Will his work on Octave--which by all rights should be able to serve as a strong-enough resume by itself to justify his hiring--even be looked at by potential employers? And what about his age? 25 years on Octave could mean he's pushing 50. I could easily see him getting a "no hire" from plenty of trendy tech companies.

[1] http://www.businessinsider.com/boxnet-2011-9

View on HN

I don't understand how YC can fund non-profits like VotePlz and the ACLU while not funding stuff like this.

Whenever I read something that starts off along the lines of "I don't understand" or "It makes no sense" I wonder how much time the poster has taken to put themselves in the shoes of those they can't understand and rough out ideas that might make sense to them.

Look at VotePlz and the ACLU, the other startups YC supports, and Octave. Do you see any similarities? Any differences? The first that jumps out at me is that YC doesn't support open source for open sources' sake. It supports organizations that serve some sort of purpose other than writing software; another way to describe this is that the organizations have business models. Some of the startups might be developing services that support software development or tech rather than something consumer facing, but it's the service, rather than the software, that is being supported by YC. In that sense, VotePlz and the ACLU are similar to the startups YC supports. Octave isn't like this. What is Octave's business model, other than to develop software that can be used by others?

Please don't get me wrong. I see a lot of value in open source projects, and have used and contributed to open source projects both professionally and personally. I don't have any experience using Octave, but have no issue giving it the benefit of the doubt that it's a worthwhile project.

Perhaps I'm mistaken. Are there software projects similar to Octave that YC supports? Is there a meaningful difference between YC's startups and VotePlz and the ACLU, and open source projects such as Octave?

Put that way, YC's politically leaning activities remind me of lobbying. Supporting policies that are good for its business models.

They could set up a fund of sorts and just take stock of what FLOSS software the current batch/all running batches use and distribute some amount of money that way.

They could also just hire prominent FLOSS folks as exclusive consultants for their batch companies which could actually turn out to be rather valuable.

They could also set up a special mini-incubator focused on helping FLOSS developers build a business around their software.

I think there are a lot of rather interesting strategic options.

I think that is a pretty great idea.. essentially a private tax system to pay open source. However like all tax systems, it would only work if there were an enforcement mechanism.

I do find it ironic that many of the dev crowd seem to love government taxes yet when it comes down to actually putting money into the products (code,) they depend on, people seem to get free-ridery all the sudden. Yet people have no problem wanting to tax people for whatever government program of the moment is en vogue.

I don't use public transport (I live in the countryside in France,) but I do use open source every day. I pay taxes for subsidizing the Paris metro (via national taxes,) yet I never (or very rarely) use it. Every person using a smartphone on a train is using open source in some form.

My logic isn't well formed but my point is that open source is every bit as "worthy" as a bullet train from LA to San Fran -- and probably benefits far more people. However since open source doesn't result in certain politicians' associates getting rich off of land deal, open source doesn't have much political interest.

To be clear, I am not proposing a government tax to fund software -- I am suggesting that investors/devs/companies that use open source ought to consider ways to support the tools that make their business possible.

Free is never free -- some dude (or lady) in a basement somewhere is forgoing personal income to help build the tools upon which billion dollar companies depend.

Open source is similar to someone personally doing the work to build a road -- a road traveled by companies making money by using that volunatarily created road.

> I do find it ironic that many of the dev crowd seem to love government taxes

Can you provide evidence for that?

Will his work on Octave--which by all rights should be able to serve as a strong-enough resume by itself to justify his hiring--even be looked at by potential employers?

It'll probably register on the backs of their eyeballs somewhere. Some of them might even might even experience some momentary flash of recognition: "Octave? Right, yeah. Some GNU thingy..." before catching themselves and saying "Well that's, nice. But he doesn't have any Node or Angular, and I don't see any of the buzzwords, like 'microservices', 'serverless' or 'container' I was hoping so far. Smart guy, but too academic. And probably too fossilized at this point in his career to learn any of the shiny new stuff... but what the heck, let's give him a chance -- we'll see how he does on HackerRank".

Ha, I'd love to see a youtube series in which famous old-school open source devs apply for web front-end gigs. Episode I, Linus Pair Programs.

Funniest post on HN in years!

This reminds me of the Joshua Bell-on-a-subway project some years back. :)

: read en-dash

God that'd be amazing. Should try to make it a thing.

You mean like Google refused to hire Homebrew's author, because he failed to make an inverted tree exercise?

> Will he be whiteboarded?

I like the way you word it - it makes it sound like a form of torture.

The lights in the conference room flickered intermittently, a slight buzz heard each time they dimmed. Something was dripping on to the ceiling tiles from above, a large orange colored stain had developed in the corner.

The interviewer leaned in again, his tie loose, his face covered in stubble. He'd been at this for four hours now and it was showing.

"Tell us your social media passwords."

The candidate flinched back, shaking his head.

"I told you, I...that's illegal, you can't ask for those."

The interviewer clasped his hands together, cracked his knuckles, and flashed a big grin. With a swift motion he swiped a marker from the table, flicked off the cap, and stabbed it in the direction of the candidate.

"Write a linked-list implementation in C. Oh, and because you've been such a tough-guy you only get to use malloc...once. In the main function."

A bead of sweat quickly ran down the candidate's face. He looked tense, fraught with concern. The sound of his teeth grinding together wasn't hard to hear.

"S-seven salamander six exclamation three five."

This story was fun to read but you lost me at the pass phrase with four numbers as words. That's clearly not dice ware!

65324 zero 44346 one 62245 two 61154 three 26252 four 25534 five 54225 six 53415 seven 24253 eight 43526 nine

Of course, if these were your dice rolls, you're pretty unlucky, and should try again. And/or buy a lottery ticket.

Someone upvote the parent, I can do it just once!

> Oh, and because you've been such a tough-guy you only get to use malloc...once. In the main function

Oh yeah? How about I just don't use malloc then:

    static char ALLTHEFUCKINGMEMORY[1024*1024*1024];
    static size_t NEXTBYTE = 0;
    void *myAlloc(size_t size) {
        void *ptr = &ALLTHEFUCKINGMEMORY[NEXTBYTE];
        NEXTBYTE += size;
        if (size & 3) NEXTBYTE = (NEXTBYTE & (~3)) + 4;
        return ptr;
    }
    void myFree(void *ptr) {
        fprintf(stderr, "REAL PROGRAMMERS DON'T NEED TO FREE MEMORY\n");
        exit(-1);
    }

nice one!

for single linked list you can just use an array of the struct and borrow/return via circular list. my C is beyond the rust levels of 20y Golf 2 but I guess I won't sweat over.

An alternative solution would be cheating away and using realloc instead of malloc.

I have a theory on why Google insists on whiteboarding senior-level candidates. I don't think it's for checking for talent as much as for checking for motivation.

Think about it - in order to pass, you basically need to drill those algorithmic exercises for weeks before the interview. If you do, it's a very good signal that you really want to work for Google (similarly, investment banks, management consultancies, prestigious law firms make their associates go through a very tough ordeal after hiring to weed out those that don't want this career bad enough). And it makes sense - for a majority of jobs at Google, you don't need to be super talented, but you need to have internal motivation to keep working on the same boring code for months/years before (if ever) you get picked to work on something interesting.

This. Because of the massive amount of hiring the big 4 do their entire interview process is available online. Sure there are variations depending on the interviewer you may end up in a room with but for the most part they all follow the same format. Anyone who really wanted to work at one of these has to study and practice how to interview for THAT company. Sure there is still some element of luck / right time right place in play but most people who fair interviews at the big four just don't put in the time required to prepare.

You're saying getting in Google has nothing to do with intelligence?

Have you gotten an offer from the Big4?

Intelligence? That's a fairly hard thing to measure and likely a poor indicator of a good employee. However there is a element of prep and memorization which does require a base level of intelligence.

Haven't gotta an offer from the Big4 but like I stated half of my former team now work at one of the big four and I've worked with loads of people who have received offers or currently work at the Big4. I also know countless people who have been utterly destroyed in technical interviews with the Big4 because they didn't put in the time and effort to study for it and treated it like a tech interview for a digital consulting agency.

Yes you can bet the farm that if he interviews at Google, Facebook, Uber, Amazon, Dropbox, etc. they will all whiteboard his ass (if he even gets an interview --because of his age). They don't give a f' who you are or what you did. In my google onsite they didn't even ask about past projects (who cares really?). The question is can you whiteboard?

> They don't give a f' who you are or what you did.

I know a few people who Google has hired into specific roles directly due to their previous work...

What she or he said is mostly true, though. But there are always exceptions.

Yeah, I can't imagine Rob Pike or Guido whiteboarding for their Google interview... Although the thought of such a conversation has me laughing internally.

I created Unix, Plan 9, UTF-8 among other things. Just go ahead and put your dry erase pen back in your pocket sonny.

The good thing about people who math often (which I presume he can since.. Octave) is that they can whiteboard.

Probably this is the right thing to do, because this is a weak-link network problem https://9clouds.com/blog/weak-link-networks/.

> I also shudder to think of how Eaton will fare on the job market should he actually be forced to seek regular employment. Will he be whiteboarded?

'Whiteboarding is a form of programming torture in which a candidate must code algorithms he will never use on the job on a whiteboard in front of other people, causing the individual to experience the sensation of inadequacy.'

You said it yourself—Octave isn't trendy or cool.

The problem is I think Octave missed its window. 20 years ago when I was in grad school, everybody was using Matlab and a competitive open source version would have been a killer app. I looked into Octave at the time, but it was basically a toy at that point. I looked into it recently, and it really has come a long way, and even has an official GUI similar to Matlab's, but these days Matlab is kind of fading out and R is what everybody seems to be using.

I did my uni ~10 years ago and I made a point of using Octave when everyone else was using Matlab. It worked just great for all the ODE and statistics exercises. Nobody else used it though.

And 10 years ago when I was in grad school (dropped out though), Python was on the rise with numpy and matplotlib. Coming from a programming background, it was an easy choice and I'm still using those today. They've got more usable for people with less programming experience with Jupyter notebooks etc.

Was briefly tempted by Octave, but it just didn't happen.

Different communities. Matlab and Octave are big among capital-E engineers, R is for statistics and fights with Python.

Exactly, and matlab is growing in engineering schools and big engineernig firms, not fading out

0 years ago when I was in grad school, everybody was still using Matlab. Ok, some used R and NumPy but 80% Matlab. Such a collection of sparse matrix routines (direct solve, iterative solve, preconditioning, eigensolvers, SVD) is very useful in applied sciences.

Or Python. I switched from Matlab to Python and Numpy and Matplotlib and Jupyter a few years ago and haven't looked back. They really are better in many ways.

20 years ago when I was in grad school I did use octave to do my research, and it was 'mostly fine'. (Also I discovered that a couple of the matlab toolboxes I needed were just .m files and there was no technical reason that I couldn't just copy them into a local directory for octave to find; I consider this no worse than 'morally iffy' considering that we had a matlab site license)

Hey, I used it in Andrew Ng's Machine Learning course on Cousera. It was that or Matlab.

Nothing trendier than ML, so I'd recommend going that direction.

You can use Python; someone created bindings:

https://github.com/mstampfer/Coursera-Stanford-ML-Python

The easiest explanation would be he just didn't ask YC.

Reply on news.ycombinator.com