utopkara

Late to the party, but here's an eclectic mix:

- Logbook of your work, plans, thoughts; use whatever tool that you know will be around the longest. Be it git, apple notes, or any other. Make sure it has a search, and keeps track of dates. - Somewhat related; an infinite terminal history. You may have the best one liners, but you don't want to keep rewriting them. Also, know your unix tools. - Computer Science; as covered in GRE computer science; those are just fundamental, if you try to find your way around learning them, you'll just waste time and effort; just go through the learning. This is assuming you are a software engineer. - Competent peers and good ways to review work, and collaborate in general. Code reviews are great, but they are just a small fraction of what you can achieve with a team. Retros, design reviews, pair coding, planning for common objectives are all productivity multipliers. - A focus on objectives; this is a basic one, but not an easy one. You'll be making many decisions when you code, design, review, discuss, communicate, argue, negotiate, etc. if you don't keep track of the various objectives, you'll waste time and effort either immediately or in the long run. In a time crunch, stop and think about your objectives, and whether you can work towards them more effectively. - Basic drawing and handwriting; you need to get comfortable with using a pen or a marker, and the ways of putting your ideas on a piece of paper or whiteboard. I've witnessed many impenetrable computer science theory or just basic design problems turn to dust when beaten down on a whiteboard by clear minds. - Regular breaks, walks and occasional coffee. Being tied up on a (hard) problem for many hours without breaks is usually a waste of time, also unhealthy for your body. - Knowing your data; if you are trying to solve a problem with ML and you don't know what your data looks like, you are wasting your time. Understand your data first. This requires experience, looking is not seeing or understanding. Also, know your objectives, see above. - Know your sources of wisdom; these could be people, books, or search tools. You are unlikely to be the first to run into the same problem. If you are looking for a novel problem or solution, it is mandatory to check what has been done before. If you are dealing with a tough problem, you might get insights from others. If you are looking for better documentation, maybe it's out there somewhere. Don't forget to check before you get lost.

The author has one opinion, here's mine:

The OLPC designers, developers, engineers, educators (and whoever contributed to the project) have done a fantastic job.

I got an OLPC for my kids through their Give 1 Get 1 program. The beautifully designed little laptop was a fantastic learning tool for children. The software and hardware kept my daughters engaged and exploring. OLPC was a way better educational tool than any available ipad software at the time. Considering the unique experience, and even now it is better than an iPad (although I am assuming the software might have caught up by now). My daughter loved programming scratch on OLPC growing up, and she preferred it to other computers and ipad too. Later on, I even installed the OLPC distro on my old Thinkpad because she wanted the simpler interface. During covid, they became more familiar with chrome and iPad, although, had it been pushed further, I believe OLPC would be a much better platform than Chrome, and iPad for learning during covid.

Depends on the people involved.

Some are likely to be sharing screen in a presentation at any given time, then it is the expected way of starting a chat.

Others would be unlikely to be in meetings often, then it is plain unnecessary (and even pretentious).

Logged in to say kudos. This is a fantastic idea. Driving around in Antalya now, and I am transported mentally.

A lot of the children's stories I heard growing up included violence of sorts. Not saying it was necessarily good, but it is very common, hence perhaps is either harmless or maybe beneficial in a convoluted way. fwiw, the stories kept me from wandering into abandoned old houses, tall thick bushes, or too far away from home; we were pretty much on our own when I was a kid.

It is hard for one to appreciate the value of a process that helps teams operate well in the long run, in the face of disruptions, pivots, failures, growth, etc. If you can do without standups, good for you.

You usually do not even want a global optimum with a high capacity model such as a DNN anyway (SVMs memorize only a small number of data points), because that possibly means that you are overfitting to your training dataset.

Meeting room switch does take time, and it is a cost, cannot argue with facts. However, real estate cost is not high only if the company is located in the woods. If one is looking to grow workforce in a central location, the real estate cost is through the roof. Keep in mind that companies do not start at their final size; so the efficiency of the space will be far lower during the earlier stages.

Given your requirements, I would suggest a consultant/contract developer position for large/established companies building enterprise solutions. Large, because they will have challenging business problems that they cannot deliver through their normal project development process. Enterprise solutions, because the problems will be better defined and your experience will be better appreciated.

I had an office (in the company which pioneered individual offices for their dedicated building for creatives many decades ago), and I have worked in open office settings. There are times when you want to think on your own, but I don't think it is impossible to achieve if you have a good pair of noise cancelling headphones. In return, you get to interact with peers more openly. Not to mention the cost savings for the company. This is absolutely a win-win. Although, the devil is in the details, and they are neither easy nor cheap: Open office settings have to be implemented with the flexibility to occasionally work from home; ample amount of meeting rooms for impromptu huddles; well designed ergonomics; absolutely relaxing decor and setup.

Fantastic story! Looking forward to following the new breed of aviation pioneers and adventures in space!

US cities are amazing for libraries as well. I don't know much about the history of libraries, but it feels as if having a nice library was a prerequisite to being called a town. I'd say even now, having a nice library is a source of pride for towns.

The problem is symmetric, and observations will match expectations if you can treat it as such.

Put yourself in the place of the interviewer, who needs to decide on the person they will work with for the next two years. Software Engineering is already hard, and your interviewers are overworked. They are assessing whether you will be able to pull your weight. Their livelihoods are in this at least as much as yours, if not more. Interviewing is no different from cases where you have limited information to decide on a critical issue, like buying a house, choosing a surgeon, voting for town mayor. You have to separate the important flaws from the minor ones, assess the risk, and pick the candidate that is the best investment.

When you are in a technical interview, you are rarely assessed for some generic software engineer ideal, but for skills that the hiring team have in mind. The chances are very low that there will be a match with all positions you apply to, unless you are also very selective about where you apply to on your end.

Many will only read the image captions and will miss this buried "info" bit:

"Conduent, in another twist, has begun competing with the start-up. The business services outsourcer, which has $6 billion in yearly revenue, introduced its own smartphone app last year. Conduent’s entry, ConnectEBT, has significantly fewer reviews and lower ratings on the Google and Apple app stores than Propel’s FreshEBT."

He could have gone up to the top of Burj Khalifa instead. But, that wouldn't have placed him on the headlines. Also, he is running for governor. As far as I can tell from recent experience, he will win.

Thank you for this table. Yet, the footprint size you measure will depend on what you are counting. We eat meat mostly for protein; and for that beef is about 10x more efficient than wheat, and 3x more efficient than lentils by weight. Also important is when you stop measuring, ie scope: irrigation, forest clearing, fertilizers, are no small damage in terms of causing climate change, reducing carbon sequestering ability of nature. The point of taxing is to make long term invisible costs visible; if anything, industrial agriculture has been disastrous for environment; if the goal is to keep this damage in check, singling out beef as a target for taxing doesn’t make any sense.

OK sure, do we also put carbon tax on wheat, and other vegetables? ~750 million tons of wheat is produced yearly, vs ~260 million tons of beef. Excepting the case of perennial plants, unused parts of vegetables (e.g. wheat straw) have a hefty carbon footprint. BTW, rice is a case in itself.

Thank you for pointing this out. I heard about this series but never watched it, and didn't know the plot. Amazing how the scifi writers leapfrog their times in understanding of science and its implications.

Fantastic thought experiment. Not sure if it was already covered in a scifi story before, my scifi literature is pretty weak. If not, somebody should turn this into one.

This points to a fundamental problem in the understanding of planetary systems. Planets do not come to being out of nothing; enourmous amounts of matter have to come together to make up a planet. Great example of experts ignoring "details" outside their expertise.

If Steve Jobs hasn't done it, then who am I to waste money on this.

Reading a random person rant about scientifying stuff, should perhaps be evenly distributed across siblings; until you realize that writing style does give away a lot about the person writing it. You can write a decently performing classifier that predicts gender (and several other demographics) from stuff they write, even from tweets, and chats. It would be stupid to think that people do not pick up the demographic clues in writing, and set their preferences accordingly.

Somebody has taken the snob of a barista really seriously. Really nice article, regardless of the motivation :-)

Anybody who is working with a production NLP system would confirm that language is very much alive and changes over time, making it necessary to adjust your models. The meanings of words shift around, spellings change, words disappear, and appear.

Why does he have to be in the rocket? He doesn’t believe in cameras either?

Yield is a real mass production problem, not an imaginary one. You can verify it with your local manufacturer. For new products, new facilities, and new production methods scaling up the production often requires engineering breakthroughs.

The newswire is the distributor of the press release. It is equivalent to Tesla emailing all major news outlets; just more convenient and reliable. In short, Tesla is the source.

Thanks for the heads up. I guess it is other coffee shops for me from now on.

This is a great result, but it is also old school machine learning (not that there is anything wrong with that).

The machine learning underpinnings (supervised principle components) from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC387275/#s4 which is from

Hastie T, Tibshirani R, Friedman J. New York: Springer-Verlag; 2001b. The elements of statistical learning: Data mining, inference and prediction.552 http://statweb.stanford.edu/~tibs/ElemStatLearn/printings/ES...

Congrats! The regex translation dataset generation is a great idea! 10000 lines could be the MNIST for regex :-)

Pun Maths: "Mattress multiplication"