The Linux Kernel, not just because of the end product. I find it's organizational structure and size (both in code and contributors) inspirational.

Firefox. Even if you don't use it as your main browser, the sheer amount of work Mozilla has contributed to the world is amazing.

OpenTTD. I liked the original game, and 25 years after release some devs are still actively maintaining an open source clone with support for mods.

Git. Without it, it would not just be harder working on your own source code, it would also be harder to try out other people's projects.

FZF is possibly my favorite command line tool.

Kitty has recently become my favorite terminal.

My favorite thing open source has brought forth though is a certain mindset, which in the last decade can be felt most heavily in the fact that:

1. Scientific papers with accompanying GitHub urls, especially when it comes to AI. Cutting edge research is one git clone away.

2. There are so many open hardware projects. From raspberry pi to 3d printers to laser cutters, being a "maker" suddenly became a mainstream hobby.

So, continuing the story, in reverse order, on the warship and its domain setup...
One day, the CO told me that we needed to set up a proper "network". Until now, the "network" was just an old Telcom switch, and an online HDD. No DHCP, no nothing. The computers dropped to the default 169.254.0.0/16 link local block of addresses, the HDD was open to all, cute stuff. I do some research and present to him a few options. To start things off, and to show them that a proper setup is better and more functional, I set up a linux server on one old PC.
The CO is reluctant to approve of the money needed (as I have written before, budget constraints in the military is the stuff of nightmares, people there expect proper setups with two toothpicks and a rubber band). So, I employ the very principles I learned from the holy book Bastard Operator From Hell: terrorizing with intimidating-looking things. I show him the linux server, green letters over black font, ngrep -x running (it spooks many people to be shown that). After some techno-babble I got approval for a proper rack server and new PCs. Then came the hard part: convincing him to ditch the old Telcom switch in favour of a new CISCO Catalyst one.
Three hours of non-stop barrage. Long papers of NATO specifications on security standards. Subliminal threats on security compromises. God, I never knew I would have to stoop so low. How little did I know that after that...
Came the horrors of user support.
Moral of the story: an old greek saying says "even a saint needs terrorizing". Keep that in mind.

PM in daily: your turn. what have you done yesterday?

me: so i finished my PR for feature x and now i'm only waiting for review feedback there, so i can close this ticket today if no major rework is required-

PM: this is not what i asked, i don't want to know what you did, i want to know what was done.

me: uhh... okay, also i started working on task x
[note: task x, a task per definition involving a large amount of research, was very coarsly defined and it wasn't even clear to the PM what he exactly expects from me, and we agreed that the scope needs to be refined in the process],
so as a first step, i started doing some general investigations to get an overview of the topic and learn about concepts a and b-

PM: again, i don't want to know what you did, i want to know what was done.

me: okay well, i have DONE basic research on topic xy and collected information-

PM: this still does not answer my question, what's the deliverable?

me: ...so uhhh.... i read papers? i researched info online and collected and prepared information and links in a presentation which i'm also planning to present to the team-

PM: okay, can you please split your jira task in subtasks so everyone knows exactly what you're working on? otherwise we have no idea what you're doing.

for fuck's sake, shut up. just shut up

I really hate how people see Bachelor of Computer Science as a meme.

I think it's because everyone who has a Computer calls himself a IT-Expert. So today on Xmas family joined for a meal, we talked about work and study and stuff...

Auntie: "So why do you need to study to Operate a Computer, my Son built one by himself without studying, you should study something solid like economics"

First I thought this was a meme, later I found out she was dead serious.

All the math, physics, scientific research, papers, software engineering just to operate a computer 😂

I'm not gonna call her names, because it's Xmas so:

father forgive them, because they don't know what the fuck they're talking about.

Admin work, because its all manual:

- Each new project has to fill out an Excel tab in a workbook, with a list of all the major tasks and who is responsible. This then needs to be used to create a Gantt chart, manually, in the same tab, showing in what month a task starts and ends.

- Every month we have to manually enter status updates into a powerpoint slide on a shared deck. Which has a collision at least once per month.

- Once a quarter we need to do something similar as the powerpoint slides, but into a word doc instead.

- Once a week we need to track our time on projects in a tool that can't be integrated with (no API or anything). Meaning we can't link up a ticket tracking system to it, so again, all manual.

- Once every 6 months a new round of research funding opens up and we write proposals. The status for which are tracked in another Excel spreadsheet, manually, once a week until the deadline.

- The instructions for what to do with the proposals are so vague and badly documented that there is an unwritten rule, that for the first time you will have to ask a bunch of questions to the project manager. This is accepted by everyone and its just the done thing.

- Everything is stored in a dropbox style system, which has become so cluttered I can only find resources by saving the links sent out previously.

- Some of these updates / reports also get a 1 hour meeting for everyone to stand up and read out what they've entered.

- From time to time random things will need to be reported on to the higher ups (how many publications, research papers, patents, times and dates etc.). Again rather than a tool, a new Excel spreadsheet is whipped up and emailed to everyone on the team. Whoever sent it out, then has to merge the 20+ copies into 1 doc.

- Some of the staff (mostly the devs), use a ticket tracking system to keep track of everything. Management refuse to use it to track the things they need. Instead we have to copy paste from it into the word docs, powerpoint, excel etc.

- By far the most annoying. Management force all the above as they need the info for finance, accounting, legal etc etc. So we have to do it, but whenever there is a question from legal, management send the question to us. So despite having documented every facet of everything imaginable, it all gets ignored in favour of endless emails.

I once tried to to put an end to all of this madness by proposing the use of a ticket tracking system, and then building reporting tools on top of it.

... I was told that it "wasn't appropriate". Still don't know what that means.

Prof: "Hey, you can take a look at the source code that we used last year in this research paper"
Me :(surprise because other papers usually don't share source code), "Okay"

A few weeks later:
Me: "Prof, if you use method A instead of method B, you can get better performance by 20%. Here's the link"
Prof:"The source link that you mentioned is for another instrument, not GPU"
Me:"Yeah, but I tested in on GPU and I found it is also applied in my device"
Prof:"That's interesting."
-----------------------------------------------------
This is why folks, sharing the source code that you used in scientific papers is important.

At work, an idiot who has never worked on machine learning before and understands nothing about it: "You know what, machine learning isn't actually hard. It is just basic statistics and then you download the model online and that's it! There's nothing else you are doing!"

STFU, you moron! Do you think just any model can work for your use case? Do you fucking think it is easy to come up with new architecture for a very specific use case and test it accurately? Do you think it's easy to effectively train a model and do hyper-parameter optimization?Do you fucking think it is easy to retrieve the right data for your use case? Do you think it's easy to keep up with research papers on arXiv being released daily? Do you think this is fucking javascript and there's a framework for everything? Stfu!
Honestly, i hate ignorant morons who generalize stuff they don't clearly understand.

I freaking hate school curriculums, I have only 5 classes this semester, only two of them are career related and one is about QA, I want to focus on my code, projects, things that really matter but instead I have to do Socials homework, learn how to do research papers, learn about marketing. I really hate this. Besides the classes I have on development are extremely light. And I have to pay for this crap

A teacher asked for my help in some machine learning project, I told her I don't have a background in ML.

She was working on an application that classified research papers according to the subject.

I said, seems like a basic NER project, but maybe I'm wrong I haven't worked on any ML projects before. But I do have experience with web, let me know if you need help in that.

She says, ML is also web, it's just like semantic web.

The reason I stick around at my current job is thanks to a mentor who has helped me reach greater potential.

He's our senior architect.

It began with him simply bouncing ideas off me. I was a rubber duck basically. After a while I began to understand these ideas. All sorts of design patterns, cache invalidation problems and solutions, and so much more.

It was almost as if through osmosis that I began to research things and learn more and more about topics I had only barely seen in high-level articles and papers.

Once I began to contribute to the discussion, he helped foster that. I went from being a rubber duck to a protege.

My pay here isn't what it should be. The problems we're faced with are stressful and often times wear me out. I stay because I'm self-taught and I yearn for learning as I always have.

This isn't just my job, but my passion. I love what I do, and I get up happy to come here every day knowing I'll learn something new while doing what I love.

Why do most people think that machine learning is the answer to their poor business decisions. I have recently had a client who won't stop talking about how his business will grow to Google's scale if I get the model to 97% accuracy . Regardless to say his data is noisy and unstructured. I have tried to explain to him that data cleansing is more important and will take most of the time but he only seems to care about the accuracy and how he is losing investors because I haven't reached that accuracy. This is fucking putting alot of pressure on me and it's not becoming fun anymore. I can only hope he achieves his ambitions if I ever get that accuracy (Ps: From the research papers I have read on that problem, the highest accuracy a model has ever got to it 90%)

found these goodies for free today. guess I'm not sleeping this weekend.

Is it normal that I enjoy reading research papers XD ???

Soo my dad has a food printer he uses to print edible images on cakes our customers order. The food printer needs to run at least once a week (regularly) to kinda guarantee not to get fucked up with its ink, as that can damage the printer when it's dry. My dad though doesn't have regular orders...
The printer has a standard function to test all colors.
My dad asked me how this task could be managed regularly, as I'm the IT guy 🙄. His idea was to log all the dates on paper.

Now I'm trying to automate this task via Windows so we don't have to care about papers to manually log when the next test must run. On Windows the printer settings can be accessed to run this color check.

... I've got a feeling this will be another one of those tasks that I will overengineer over the top😅. I've already done my research with automated batch jobs (never done batch before) but the normally proposed code for a "Düsentestmuster", so the color check, prints a different overview I was not expecting, which doesn't fit the purpose.
Now I'm here and, as I currently see no way of simplifying it, I have to kinda simulate a person that opens these settings and runs this check. With Python, pyautogui and Tesseract OCR, to prevent the program from clicking anything wrong. Although I'm sure there should be an easier way for this, I haven't found it, so I guess I have to proceed on this path and take the experience I gain as a bonus...

Add your fucking requirements.txt files or atleast have a decent fucking readme. I have had to implement many official repos for AI ML papers now and most of them are shit with a load of bugs in them, cant you implement them properly while you are at it?

Also would like to add most of the sota results arent honest maybe famous ones are but the research community is full of shit really. All of that cant be changed but atleast add readmes and requirements ffs.
I have to spend days just to implement your sub par result providing fuckery.
I’d rather just code it all myself sometimes

INTERVIEW. It tells everything about the company. I recently applied for a "big" company for the position of ML Engineer. The Job description was like "someone with good knowledge of visual recognition, deep learning, advanced ML stuff, etc." I thought great, I might be a good fit. A guy called me the next day. Introduced himself as a manager of the Data Science team with 8+ years of experience. Started the talk saying "it is just an informal intro". But things escalated very quickly. Started shooting Data Science questions. He was asking questions in a very bookish way. Tells me to recite formulas (like big formulas). When I explained to him a concept, he was not understanding anything. Wanted a very bookish answer. I quickly realized I know more about ML stuff than him (not a big deal) and he is arrogant as fuck (not accepting my answers). Plus, he has no knowledge about Deep Learning. At the very end, he tells me "man, you need to clear up your fundamentals". WTH??? My fundamentals. Okay, I am not Einstein or Hinton, but I know I was answering things correctly. I have read books and research papers and blogs and all. When I don't know about things, I tell straight away. I don't cook answers. So the "interview" ended. I searched that man on LinkedIn. Got to know he teaches college students Data Science and ML. For a fee of 50,000 INR. It's a big amount!! Considering the things he teaches. You can find the same stuff (with far higher quality) free of cost (on Coursera, Udacity, YouTube, free books, what not). He is a cheater. He is making fool of college students. That is why I sometimes hate "experience". 8+ years of exp and he is such an a**hole!! BTW, I thanked God for saving me from that company. Can't imagine such an arrogant boss.

TLDR: Be vigilant during interviews. It tells a lot about the company.

This is gonna be a long post, and inevitably DR will mutilate my line breaks, so bear with me.
Also I cut out a bunch because the length was overlimit, so I'll post the second half later.

I'm annoyed because it appears the current stablediffusion trend has thrown the baby out with the bath water. I'll explain that in a moment.

As you all know I like to make extraordinary claims with little proof, sometimes
for shits and giggles, and sometimes because I'm just delusional apparently.

One of my legit 'claims to fame' is, on the theoretical level, I predicted
most of the developments in AI over the last 10+ years, down to key insights.
I've never had the math background for it, but I understood the ideas I
was working with at a conceptual level. Part of this flowed from powering
through literal (god I hate that word) hundreds of research papers a year, because I'm an obsessive like that. And I had to power through them, because
a lot of the technical low-level details were beyond my reach, but architecturally
I started to see a lot of patterns, and begin to grasp the general thrust
of where research and development *needed* to go.

In any case, I'm looking at stablediffusion and what occurs to me is that we've almost entirely thrown out GANs. As some or most of you may know, a GAN is
where networks compete, one to generate outputs that look real, another
to discern which is real, and by the process of competition, improve the ability
to generate a convincing fake, and to discern one. Imagine a self-sharpening knife and you get the idea.

Well, when we went to the diffusion method, upscaling noise (essentially a form of controlled pareidolia using autoencoders over seq2seq models) we threw out
GANs.

We also threw out online learning. The models only grow on the backend.
This doesn't help anyone but those corporations that have massive funding
to create and train models. They get to decide how the models 'think', what their
biases are, and what topics or subjects they cover. This is no good long run,
but thats more of an ideological argument. Thats not the real problem.
The problem is they've once again gimped the research, chosen a suboptimal
trap for the direction of development.

What interested me early on in the lottery ticket theory was the implications.
The lottery ticket theory says that, part of the reason *some* RANDOM initializations of a network train/predict better than others, is essentially
down to a small pool of subgraphs that happened, by pure luck, to chance on
initialization that just so happened to be the right 'lottery numbers' as it were, for training quickly.

The first implication of this, is that the bigger a network therefore, the greater the chance of these lucky subgraphs occurring. Whether the density grows
faster than the density of the 'unlucky' or average subgraphs, is another matter.
From this though, they realized what they could do was search out these subgraphs, and prune many of the worst or average performing neighbor graphs, without meaningful loss in model performance. Essentially they could *shrink down* things like chatGPT and BERT.

The second implication was more sublte and overlooked, and still is.
The existence of lucky subnetworks might suggest nothing additional--In which case the implication is that *any* subnet could *technically*, by transfer learning, be 'lucky' and train fast or be particularly good for some unknown task.

INSTEAD however, what has happened is we haven't really seen that. What this means is actually pretty startling. It has two possible implications, either of which will have significant outcomes on the research sooner or later:
1. there is an 'island' of network size, beyond what we've currently achieved,
where networks that are currently state of the3 art at some things, rapidly converge to state-of-the-art *generalists* in nearly *all* task, regardless of input. What this would look like at first, is a gradual drop off in gains of the current approach, characterized as a potential new "ai winter", or a "limit to the current approach", which wouldn't actually be the limit, but a saddle point in its utility across domains and its intelligence (for some measure and definition of 'intelligence').

One problem for CS education is that the salaries for academia are so low compared to industry that if someone is even vaguely competent, they can at least double their income by working a 'real' job. Now this may be different at higher levels of colleges but generally those folks are such bullshitters they wouldn't last outside of academia.

As what to improve?
Depends if it's a research or vocational course.

For vocational; heavy on group projects, common tools, methodologies and architectures. All demonstrated in something like c#/java/python. And one project must have a web app (db, app layer and JavaScript from end)
Basic knowledge of algorithms, runtime analysis (O notation) and some data structures and you're an instahire.

For research, go wild. Deep dive into the math, algorithmic side. Read up lots of research papers. Try out different programming paradigms. You would aim for a career in academia, AI, quant finance etc...

IEEE is full of shit! Why the fuck can't i download research papers when I have already paid for the subscription???
Fuckers, i didn't pay to view your old looking website!

the only way I could proceed further now is studying research papers!

I have never regretted over anything in my life like I am regretting taking a course named Speech Technology this semester.
Professor is nice but problem is he is too ambitious. He wants us to study a research paper on our own and implement whatever it is in research paper. People have worked years to write those research papers. How we gonna do it in one or 1.5 months. And the worst part is there are not much technical resources and documentation available to understand how to implement those things. Just abstract theories.

May the force be with me!

How do you feel about ieee and other paid research websites?

Every time i search something complex, an ieee research paper would pop up and i couldn't read it, coz i don't have the membership. Even if i did, i had to pay Rs. 1000 (~=$12) . For every paper i want to see

I am not saying its bad to demand a price for your work. But i wish ieee was more like github or medium, where people could also optionally publish their content for free viewing. The cost is making a lot of students miss deep knowledge of research papers.

The main thing that currently frustrates me as a student is the fact that University subject syllabus are made by sulky old phd teachers who have been long term members of ieee and other paid research orgs, and thus have designed the syllabus with topics which are covered nowhere but in research papers.

I also know that some of you sre thinking "dude , just google search anything and you will find tons of videos and content on anything", but from what i have observed, free internet takes time to grow for a perticular topic . If i search a relatively complex topic i may find some surface info and basic videos, but to go deep, i have to rely on paid/pirated books and papers.

These organisation has gathered a lot of content and renowned people. Maybe they can give away a few knowledge to the open source.

TL;DR: Have you ever been on a serious company where you have to DRAW a high fidelity mockup of the software in the design phase?

So I'm in my last year of college and I have a class called Interactive systems design, which is basically about usability and how to design the frontend of your app so it's intuitive, pretty and easy to use.

So we work in groups to design a project for the entire semester, following a long and tedious process of research and planning which includes writing absurdly long documents, doing interviews with potential users and more.
Now that we've done all of that, the teacher insists that we make paper mockups of our app before we do a digital one using Balsamiq or other programs. He wants the paper mockups to be "interactive", so we have to draw them and then record a video where someone "clicks" on the mockup with their fingers and another person moves the papers around to make it look like an actual app that's doing something.

The teacher still insists this is something almost every company does when designing a project, so it's very important that we learn to do this kind of stuff. He's kidding, right? Have you guys ever drawn a mockup of an app instead of using some mockup software?

From the last 3 years, i have accumulated interest and experience in android dev. Not sure about the future, but that's probably where i will be.
But this fact is moot to our 50 year old grumpy professors teaching 1000 year old rusted computer syllabus, who rejected my idea of a video streaming app as major project, simply because i projected it as a social media app, and "everyone is making a social media app, its such an old topic". yeah right sir, its younger than your daughter that fucks in the lobby

Now we are doing a project on file conversions website, a project suggested by my team member and my good friend. its such a shitty topic, there is no resources available, even the research papers are bad , every search points to a shitty site, and i don't know shit about web dev.

Technically i am the team leader, but my team mate won't let me make the project as android native app, because "Brooo, i am going to make a react app that would be completely offline, completely client side, full secure and shitt small" and sometimes "Bro its my idea" .

Well, 1. the whole point of client side is stupid because the 18 mb jsfile isn't going to get downloaded first in the client's cache(or whatever the process is, idk). The top stack overflow answers i saw told me to buy an ec2 instance and run liberoffice commands on it for every request, and that's SERVER SIDE. even if we could, i am sure its going to be bigger than what i would have made in kotlin.
2. what am i supposed to do? look at you coding while make all the ppts and research paper? you are going to use undocumented libs that "just works" , and i am suppose to curate the theory behind this, looking at all the researches of the world?well i guess okay that's a light job since THERE AREN'T ANY.

And we are targetting all types of conversions, nice. from what i know, handbrake.fr: video conversion s/w = 16 mb. photoshop: image conversion s/w=1gb and ms word: doc to pdf/other formats= 500mb.
Plus all those proprietary and undocumented formats, ugh. Thank you ugly ass companies.

Internet is great but web dev has become a whole lot mess. "I am going to build a software that is going to run in your system only using your device's processor" is a desktop/mobile app, not a website

Has anyone read any research papers looking into the application of differential cryptanalysis to blockchains and mining, and if so, could you point me in the right direction?

I'm bored and looking for something interesting to read on my day off. More importantly I'm ignorant enough of the subject to entertain the possibility that there might be some possible (if not plausible) cross-pollination of ideas between the two domains.

How many of you have formalized knowledge in computer science theory? Do you find yourself using that knowledge in your daily engineering life? For example, knowing random search algorithms, or obscure data structures. I ask this because of the modern "technical interview" trending towards discrete math instead of actual programming ability. Instead of coding projects I care about or reading research papers, I'm just doing discrete math problems to prep for recruiting. While it's not the worst thing to do I just wish there was a more direct way of interviewing a person's engineering abilities.

5 am seems like the perfect time to compare deep generative models and deep reinforcement, just because you can't fall asleep

Soon I will begin second year at my uni. So i have to start preparing my enginieering project. I already know what i want to do. But before i will be able to make it i need materials and tools. (I dont want money from uni cuz they will have rights to it, or so i think) my first step is to make myself a welder then make, i repeat make a lathe and a milling machine. BECAUSE I CAN. It pains me that most of the research papers are shit and practicly useles for new students so im planning on creating something that already exists but in a simple, professional way so other students can learn basics of creating something in practical world. A lot of scienctist go and push boundaries of science without caring about new people that are left alone to learn the basics. I shall correct that.

How do you guys keep up with new concepts and tech? Or what is your preferred source for this activity?

Conferences(in person or youtube)
ACM journals/SIGs
Research papers
Blogs/hackernews or the like
Books
Something else
?

!rant

I am building a side project to build a recommendation system for research papers. Since it's my first attempt, any tips? Also, my plan is to get it in the cloud, build a separate webpage to summon data from it. Honestly, most of it is three-fourths imagination, one-fourth google queries. The cloud idea is solely to provide it means to train real time.

I plan on using Python for this, other than the html/css frontend.

As someone who is very new to such a project, what should I know before I start?

Thanks,
S.

Having full access to online papers opens up a lot of benefits for students:
Online essays are available whenever you need them. You just need to have access to the Internet!
Our essays are written in accordance with all writing standards. You shouldn’t be worried about grammar mistakes because we add samples of reputable professionals to our base. It also means that structure, sequence, and logic of writing are clear, precise and understandable.
Our website contains an huge assortment of newspapers on topics that are various, which means you will not have trouble locating the one for your requirements.
You may download and use our samples! This benefit is helpful since you may look through our essays and also make up your papers that are private to fulfill both your own teachers' needs and get outstanding grades.
Specialized language can be found by you in our essays, which is quite beneficial to understand while drafting missions that are new. You could paraphrase some passages to make them seem brand new and fresh.
You can even cite some of them because quotations are used widely in our essays. Moreover, you can use references from essays to broaden your base of references. Thus, your essay may look more solid and convincing, and newly found references may help you with further research and writing.
Our site is easily navigated. It has an essay helper that facilitates to find a proper paper manyfold. It will not take much time to find a paper you really need.

Nevertheless, students who are under pressure due to lack of time and control can place an order for custom and professionally written essays. Our trustworthy service has already assisted many times in writing original and qualitative assignments. High standards of our works have helped many students to get the best grades! Your professors will be satisfied, and you will get more time to do things you really adore doing!

!rant.
Any awesome ios project ideas for the weekend? Tired of reading research papers.

Does anyone know of some good research papers around NLP AI capable of situational awareness? General conversational awareness... i remember trying cleverbot a few years back and it left much to be desired. Even Google is sub-par when it comes to asking follow-up questions.