ricksharp

For the nonsense answers, I have to say that if someone asked me that, I might very well reply with made up nonsense.

Sounds like a perfectly valid Dr. Seuss answer to a Dr. Seuss question.

To me this is more human seeming and demonstrates more personality then just saying, “I don’t understand your question?”

Of course as others have pointed out, if you want it to get serious, you have to demonstrate that it should respond with a “That’s a Stupid Question” if you want it to do so.

What would be interesting is if could master the usage of a sarcastic or silly emoji to tag when it was making stuff up versus being serious.

Yeah, I saw a text => UI generator.

It’s cool, but it looked like very basic stuff - the type of UI that is very easy to create in a few minutes. (And really with what was setup behind the scenes - maybe just as fast to just write the code.)

The hard part about software development is not those bits which are common, but the parts that are unique to our specific solution.

Those types of contradictions are what made me suspect the article was generated.

Now, I’m not so sure :)

With bot-human cooperative text generation this could be very powerful (if the purpose is to generate a large volume of text).

I could tell that something was off when reading through this. The logic didn’t flow and there were contradictions that defeated the previous point. Even more, it wasn’t clear if there was a theme - more like a rambling story.

However, I wouldn’t be surprised if someone told me this was written in a High School English class - as a rough draft.

That is where it occurred to me, that if this was given to a human editor to clean up, it could be interesting.

For example, if writing a creative work of fiction - this could generate a starting point for the human author to refine.

This could be done in an interative process:

- Generate a chunk of text (or multiple chunks). - Edit that into a final text. - Use that as input for the next chunk. - etc.

Right, I’ve been reading about machine learning for about 5 years, have read hundreds of articles about different techniques, and have often tried to explore ways it could be used.

However, I’ve never found a practical use in software engineering.

Every time I think I discover something that could use machine learning, I usually don’t have any data to work with or don’t have a clear definition of what the inputs and outputs would be.

In the end, I find a way to develop a solution that doesn’t need AI and often makes me realize that AI would not have been able to provide the required reliability.

I even participated in a self-driving RC car competition (running on raspi 3). I ended up scrapping the ML solution because it could only run at 10 hertz which was not even close to what was needed.

I ended up writing a custom algorithm in C that ran at 200fps and could do “advanced” moves like backing up and course correcting.

I’m not saying that ML doesn’t have its place. I’m saying that right now it’s all about images and nlp and has huge costs.

In any practical usage, the ML parts that I would need are already available as a service that I can use almost free (like azure cognitive services).

Each of these projects is something I think I could probably figure out how to make in a year (but not really - I simply couldn't do the hardware side with my toolset).

Then I look at his youtube videos and see he is putting out one of these videos each week.

The level of productivity is insane.

For example, in the basketball 2.0, he mentioned doing the 1.0 version a few weeks before. Then in a few weeks, he is doing something equivalent to Mark Rober's dart board (that took Mark 2 years with help) - while creating other videos.

How is this possible?

Wow, amazing work - the skillset includes everything from machining parts (Using a plasma cutter) to custom 3D software (for simulating and debugging).

My point was that if testing for a skillset, using something that favors those who have toyed with a specific algorithm is the same as asking trivia questions.

Trivia does not provide evidence of skill, it demonstrates prior knowledge.

It would be better to introduce a unique situation that would place everyone at the same starting point.

In fact now that I think about it, it might be a good test to see if they are passionate enough to be a great problem solver:

When was the last time you wrote a sorting algorithm?

Acceptable answers:

- You mean sorting a list? - In school - Why would I write a sorting algorithm? - Never - (Glazing eyes and other body language that indicates frustration)

Red Flags:

- Anything that indicates they think writing a sorting algorithm is a perfectly logical thing to do during the course of software development.

(Note: The problem is sorting may seem like a well-understood problem, but it’s actually not. Only a person who studied the algorithms in school as an assignment would feel comfortable with that code.)

This sounds like an inspection of whether the person has had a class in basic algorithms rather than if they have ever coded anything in real life.

In school I played with sorting algorithms, in business if I ever found a developer manually writing a sorting algorithm, I would consider them inept (unless there were very specific reasons to do so).

If someone didn’t know how to sort a list using the built in or standard library of whatever language they are using, I would know very quickly that they have almost no experience writing anything useful.

Nice work!

I’ve created my own blog around a pure pre-render solution (using gatsby stripped down to just pre-rendering) - because I want to control everything and not be dependent on anything aside from the stack (react in my case).

Impressive how you’ve tied everything in to github. I’ll be reading your code to see how it all works.

Ok, so I am just trying to understand the basic concepts in the paper and put it in my own words:

It seems that the primary idea is that quantization precision is more important where there is a high density of neighbors.

I.e. at the edges the quantized sections (buckets) could be large since there are few items there, but at high density areas, the buckets should be much smaller in order to have an even distribution of objects per bucket as possible.

Therefore, the overall effectiveness of a Quantization loss function should not be evaluated on a sum of squared error (that assumes the vector space has consistent linear value), but should rather consider the densities of the vector space and use that as a weight of the errors at different regions.

To me it seems analogous to a hash set, where the goal would be to have even distribution (same number of items in every bucket).

We want to quantize space so that every position has about the same number of items.

The mistake is creating an abstraction because of seeing duplication.

DRY is not a good guiding principle. It is an anti-principle.

Abstractions should only be created when they have a clear purpose and create a simpler architecture by encapsulating a single concern.

The reality is that all code is duplication. The reason we write code is because it is the most concise language to specify the intended goal in the current context.

What is unique is not the code that we are writing. The unique part is the code in the current context and each level of abstraction separates the context from the implementation - so that abstraction must be beneficial in organizing the overall solution into individual logical components of singular concern.

I’ve often wondered if nightmares serve to prepare us to act quickly in worse-case type scenarios. Our mind responding to stress by saying, ok if this goes really bad, what are you gonna do about it.

That’s pretty cool. It seemed a bit slow creating the “macro” but pretty cool to see it run through all the lines.

The way I would do that in vscode would be with a multi-edit, but it would get iffy at that many lines.

I‘ll record a video doing the same for comparison tomorrow.

This makes me wish there was a site for recording challenges like this where developers could race using their favorite tools.

Vim vs VScode would be an amazing way to learn tips from experts on both sides.

Thanks for the feedback.

vscode has regex find replace, but I don’t use it very often. It does show every match highlighted in the current file while entering the find value, which is nice.

It does the same for multi files if needed and gives a nice preview of every change.

Word navigation I use a great deal (especially with multi-cursor editing). Ctrl+left/right jumps words, home, end to begin end of line.

Ctrl+f Z Enter would go to next Z value. (Which I don’t use, because I don’t think in partial words much, but rather jump to mouse if I need to navigate far)

That’s probably where I could gain in using that instead of going to the mouse every time.

My first few times in linux after trying to read a file:

Ctrl+C esc esc Ctrl+c Ctrl+q

What is this! What is going on!?!

15 minutes of searching on my phone later, q...

Ok, also am curious about multi-insertion with cntr-left/right that vscode does:

For example, let’s say I have a json object with variety of field names, one per line and I want to remove the quotes.

With vscode, I can alt select all lines, Home Ctrl+right del cntl+right del.

The trick is the field names are different length, does vim have multi-edit with the ability to navigate relative to each insertion point?

Edit: Here is a good video example (https://youtu.be/iomWEDMBlcE)

Can vim do something like that?

Thank you for that example!

Just thinking out loud how to do that in vscode for comparison:

- I don’t think repeating actions a specific number of times makes sense in most use cases. Counting 10 lines would take longer than doing shift up/down n times (which would be exact with feedback) to select the desired lines.

- Multiple Copy Paste buffers sounds interesting, I might look up a vscode extension for that. (Had a copy paste queue in Visual Studio but it would break sometimes and would lose something important).

- go to start of function jnk, there are many ways to do that in vscode. Ctrl+f jnk enter would be one.

- Goto start of document is Ctrl+Home

- Indent 20 lines by 10 spaces, could do Alt+PageDown, then up/down to get to the right place (without counting or caring that it is exactly 20 lines), then with multi-line insertions could space,tab, etc.

That operation would probably justify mouse usage though, in which you simply hold alt, drag ~20 columns/~10 cells and hit space - which itself would be faster than counting 20 lines.

- find junk: ctrl+f junk enter

- select 15 lines (shift+page down,up/down to adjust) I can preview I have selected the correct area, other possibilities to select current function, etc.

- Ctrl+h to replace, a1 tab A1 Alt+L Enter

- of course in most cases, I am renaming a symbol, so “F2 A 1 Enter“ would do

I think the reality is that counts are not known before hand, so being able to increment/decrement with visual feedback is perhaps faster than a miscount, but I’d have to see it to know.

Anyway, thanks for the detailed examples! I can certainly see how it would be powerful if committed to muscle memory, but even with vscode there are many shortcuts I don’t use because of formed habits.

Does anyone have a good video comparing vim speed coding vs vscode using built in editor features?

I’ve been using vscode for years and the YouTube videos that demo vim features are shallow (like how to enter a blank line - really?). I can do the same thing I have seen in vim videos without memorizing a bunch of strange key bindings all while being able to edit code instantly without changing modes.

I would love to see a fair comparison between someone using vim at 100% and someone using vscode at 100%.

Of course some vscode shortcuts might not be fair (like F12 to jump to definition, F2 to rename every usage, etc.)

But even in raw editing:

With vscode, I can do multi-line editing, I can navigate quickly, jump to specific words, etc all very fast.

If there is no video like this, would anyone be interested in a race?

Almost every question I search these days follows this template.

It may the nature of experience.

The type of questions that cause me a problem are deeper edge cases that had later fixes for early workarounds.

I don’t understand the value of container tech (unless dealing with legacy code).

Serverless seems better and cheaper (lambda, GC Functions, Azure Functions). Pay as you go with no overhead costs. In many cases usage may never exceed the free credits.

For the database and persistent storage, you need a cloud native service like S3 and Postgres RDS. (Running a database in a container is asking for complete data loss from what I understand.) This is the primary cost for my tech stack and for lite loads $15 is about as minimal as possible.

So I don’t see how using containers could be cheaper unless you are keeping the data in the container, which is a bad idea since containers are supposed to be stateless.

So serverless cloud native tech seems better by far from everything I’ve experienced.

Also when using a tool like serverless framework, deploying an entire stack of serverless resources is dead simple.

I’ve tried a few times to see the value of containers, but I just don’t get it.

The entire reason kids and adults should write in plaintext is in order to focus on the content of their message.

Markdown is excellent because it prevents the writer from formatting except where that has semantic meaning (headers, lists, etc).

I’m teaching my son (14) how to use the best tools: The first assignment I gave him was to write a short essay in markdown, using vscode, and commit and push the answer in a pull request on github. I can do a code review, pointing out changes or even add my own content in a commit of my own.

These are powerful tools that everyone should be learning for writing intellectual content. The diff tools and being able to navigate change history lead to improved quality. In the same way that git gives me confidence to delete bad code, when writing an essay, I can delete low quality content. While rewriting I can see the diff changes for reference.

Of course, once you are ready to “deploy”, you will have a suitable style transformation so that the final result is a perfectly formatted document (with emphasis on excellent content).

TIL what TIL means (also Veblen Good). So today I learned 2 things.

I think this needs many up-votes.

It reminds me how sometimes converse is confused with contrapostive.

Your link is broken. Also does this allow for Remote or is it onsite only?

Thorn is an amazing group! Thanks for your efforts!

I think this is a great idea. I usually read at night to help fall asleep. Unfortunately my finger sometimes slips and I end up highlighting half the book in the Kindle app - which then takes me a while to figure out where I should actually be :D

I sent an email about considering a pure JavaScript (Typescript) Stack with React, Node.js, and microservices on the cloud with a Json data store. Hope to hear back soon.

Could someone list some practical examples where Differential Programming would be useful?

I am familiar where Nueral Networks and Convolutional Networks have done well especially around image processing etc.

But I can’t imagine where having differential code would help unless it is just tying multiple neural networks together in a continuous chain of differentiation.

For most programming tasks, I can’t imagine how differentiation would be possible or beneficial.

Is there a possibility that one could start with a series of unit tests and partial results and through gradient descent actually arrive at additional passing test cases? Most of the time in my experience, passing additional test cases like this requires significantly more complex structures that would not be found via differentiation.

Thanks Jade, didn’t realize I grabbed the wrong link.

This paragraph from the paper was helpful for me:

Differentiable programming is of joint interest to the machine learning and programming language communities. As deep learning models becomes more and more sophisticated, researchers have noticed that building blocks into a large neural network model is similar to using functions, and that some powerful neural network patterns are analogous to higher-order functions in functional programming [Fong et al. 2017; Olah 2015]. This is also thanks to the development of modern deep learning frameworks which make defining neural networks “very much like a regular program” [Abadi et al. 2017; LeCun 2018].