Dells XPS are made of magic. [long story, major fuckup, 10k+ damages]

It all started in December. One morning I was late to work, drove there as fast as possible. (I live like 3 minutes away so me being late really meant *late*) Parked my car in a secluded car park, grabbed my backpack and ran to work. The car park is like 100 meters away from work so I took my feet into my hands and ran. Next thing I know my heels loose all grip while I go down a small slope and I drop on my back full force. On a sharp edged stone. With only my 1700$ XPS in it. Fuck.

I paniced, but got up and ran to work. I checked on the notebook, praying it would boot. It booted! Holy shit. I flipped the notebook and saw two small dents in the aluminum shell. I was thorougly impressed. I later discovered that it left a small shadow on the display, but given what a hit that was (I am not exactly a lightweight), impressive would be a massive understatement.

Fast forward to February, I am weighing my options to get the screen replaced maybe, as damage on my hardware (even if neglectable) triggers some sort of OCD and makes me feel bad 24/7. Also my laptop tends to shut off from time to time, looked into the Event Viewer and saw kernel panic. I figured that the battery probably still took a hit and that it drops voltage from time to time and the kernel assumes a critical situation, thus shutting off.

It stayed quite snowy in Austria up until March, so occasional snowing wasn't rare. Got out of work one day, saw it snowed a bit. Whatever. I had my moms car at the time, so I tried if it would slide a bit if I donut on the now (5pm) empty parking space. Nothing. Drove done a small hill, ABS triangle lit up red (board computer can't outbalance the snow). I drove out to the main street where everything was salted and drove along towards my house. Took a turn into my street, accelerated for a bit and then went off the gas so the car would smoothly drive along with the speed slowly degrading. So I went off the gas and noticed I was a bit to the right, no wonder, centrifugal forces.

*steers left*
"Huh seems like I need a bit more"
*car still doesnt move much*
"What the- go to the left!"
*steers left hard*
"Fuck that wall is coming closer"
*Breaks*
*car doesnt break*
"FUCK FUCK FUCK FUCK!!!"

Everything got quiet in seconds, me waking up to an open airbag, ripped pants, a hurting wrist, the radio somewhere on the ground and fumes that smellt like burning wires. I grabbed my backpack that was now somewhere on the floor instead of on the seat and ran outside, tears in my eyes and the phone on my ear calling my mom. I walked inside as she walked outside, hearing a weeping scream that I haven't heard from her since I am alive. While walking inside I noticed my backpack was wet on the bottom, my 2 litre water jug shattered when my backpack hit the dashboard. I tried to stay calm and act rational, knowing that every second counts when It comes to water damage. I hastely searched for some rice and a bag to put my laptop into, stuffed the bag with both and went outside. The car was totaled, my mom pissed and crying. And I was in shock, sad, angry and hurting.

I kept the laptop on my heater for a few days, bagged in rice. I dared to try a boot after a while and you wont believe me, it fucking booted. Even the keyboard backlight worked, just the screen was obviously broken in the back (no color distortion or bad pixel rows though!!) and the aluminum shell had a dent on the front. I talked with Dell Support a few days later, asking if it would be ok to open the XPS up so I could drain all of the water. She said yes thats fine, as long as I dont touch anything or screw around with it.

She said I can send it in and get it checked, but the pickup and analysis will cost 150$ and I can go from there.

I sent it in and estimated that, because battery, screen and other things probably needed changing, it will be around 900$.

Got a call a few weeks later:

"Hello beggarboy, the repair team reported back to us and said that they will have to replace everything, which will be 1700$."

"Fuck... Buying a new one is cheaper.."

"Yeah I know I am sorry about that, I can offer you a voucher so you can buy a new one for 250$ off if you would prefer that"

"Sorry but I will need some time to consider"

"I understand."

The agent clearly noticed I was bummed about it.

After going back and forth what to do I got another call a few days later.

"Hello beggarboy, we talked a few days ago. I have good news"

"Hello, yes, speak up?"

"I was able to get a special offer for you after putting in a few words..."

The next thing she said seemed unreal to me.

She was able to cut 600$ (!!!), making the new offer 1100$, instead of 1700$ or a new one for 1500$. I figured the reason she probably did that was because I am always very polite with support members. Always.

My XPS is back and healty again.
Thank you for taking the time to read this.
Dells XPS are made of magic.

So happy i finna got a good pixel sort algo goin! Been trying for like 6 months 😃 !

A beautiful gem ticket from a manager today:

Title: "Check Stripe "Snippet APK" that might help for integration into the app to track pricing easily."

Alright, it's very clear this particular individual has no idea what they are talking about, but, I'll give them the benefit of the doubt and read the ticket description!

Description: "I think stripe offers some sort of snippet that can be implemented into the app similar to FB pixel. (I could be wrong here..) let’s briefly check this, if it’s of value for our A/B-Tests → e.g. if it makes your life easier = good otherwise it’s not important."

...

I might as well replace the management team with GPT-3 at this point.

Or even just a simple Markov chain; that'd probably be more accurate if you want to match the ticket quality more exactly of this ABSOLUTE PILE OF HORSESHIT WASTE OF TIME I GET FED EVERY SINGLE FUCKING DAY.

🤡

So I've had a fan game idea in the back of my head and have no artistic friends or money to hire an artist so I'm attempting the pixel art myself (again) and in the past 2 hours I have drawn a sort of shit looking barstool... Ugh why can't I have ideas that can survive with developer art -.-

Trying to make use of Google Maps. Search results are always sorted by "most relevant" by default, instead of by distance. Always I switch to distance, next search goes "most relevant" instead, which should be labelled "most irrelevant" instead, as it seems to be an excuse to show a list with promoted businesses not only far away, but often unrelated to my actual query. Wasn't Google supposed to be some sort of search engine experts? or at least the lesser evil of search engines? Oh wait, no, they're actually an advertising company, and it shows. Fuck you, Google. Where am I going to throw my Pixel phone and why did I open your shitty apps in the first place? I should switch MY own defaults back to Open Street Map etc.

Sus!
yesterday I bought a cool domain in namecheap, I was very lucky to find short and good one for my case.

Today (at weekends!!!!) I receive a letter:
>Hello **redacted name**,
>
>We are contacting you from the Namecheap Risk Management Team regarding your '**redacted name account**' account.
>
>Unfortunately, your Namecheap account was flagged by our fraud screening system as requiring verification and was locked.
>
>Please follow the instructions below to get your account verified:
>
>- take a color photo of the credit card used for the payment at **redacted link**
>
>Please make sure all of the edges of the credit card are visible, and that we can clearly see the card holder's name, expiration, and last four digits of the card number. The screenshots or images of the card cannot be accepted for verification. >If the submission does not meet these requirements, we can either request to submit the details again or permanently suspend your account.
>
>- provide a valid phone number and the best time to call you (within normal business hours, US Pacific time).
>
>If we do not hear back from you within 24 hours, we will be forced to cancel your orders.
>
>We apologize for any inconvenience that may result from this process. This extra verification is done for your security and to ensure that orders are legitimate. This industry, unfortunately, has a high rate of fraudulent orders, and this sort of >verification helps us drastically reduce fraud and ensure our customers remain secure. Such documents are used for verification only and are not provided to third parties in any way. Account verification is a one-time procedure, after your account >is verified, you will never face this issue again.
>
>Looking forward to your reply.
>
>---------------
>Dmitriy K.
>Risk Management
> Namecheap, Inc.

what if I did not notice it in 24 hours? It is the weekend for god's sake! People usually rest until monday.
They would what, cancel order and scalpel it to super high price?!
I have some doubts if the request is trully having anti fraudulent origins.

What if I used digital visa card? How was I supposed to photo it?

And the service they provided for photoing accepts only photos from web camera. I was lucky that I bought recently web camera with high enough amount of pixel power and manual focus. What if I did not?

That's all really SUS!
The person can not notice the letter within 24 hours time frame until the morning, when it would be already too late.

Approaching the limit of what TI-BASIC can do without busting out the modern calcs. Doing a simple operation on the whole 95x63 1bpp graph mode screen, say, turning all pixels on or off, takes over a minute. Add any sort of calculation to that...
I'm already using BASIC-callable machine code snippets to scroll the screen one pixel (which are nigh instantly finished,) and i'm so fucking tired of scrolling effects...

God, adding sound is gonna be a nightmare...

EDIT: for reference, dev machine is a TI-84+ Silver Edition. Not one of the ones with the eZ80 and backlit color screens, the greenish Z80 1bpp screened one. The one that's an 8MHz Zilog that TI decided to make multitask. The one where oscillating the screen at an integer division of full framerate fries it in seconds?

The next step for improving large language models (if not diffusion) is hot-encoding.

The idea is pretty straightforward:
Generate many prompts, or take many prompts as a training and validation set. Do partial inference, and find the intersection of best overall performance with least computation.

Then save the state of the network during partial inference, and use that for all subsequent inferences. Sort of like LoRa, but for inference, instead of fine-tuning.

Inference, after-all, is what matters. And there has to be some subset of prompt-based initializations of a network, that perform, regardless of the prompt, (generally) as well as a full inference step.

Likewise with diffusion, there likely exists some priors (based on the training data) that speed up reconstruction or lower the network loss, allowing us to substitute a 'snapshot' that has the correct distribution, without necessarily performing a full generation.

Another idea I had was 'semantic centering' instead of regional image labelling. The idea is to find some patch of an object within an image, and ask, for all such patches that belong to an object, what best describes the object? if it were a dog, what patch of the image is "most dog-like" etc. I could see it as being much closer to how the human brain quickly identifies objects by short-cuts. The size of such patches could be adjusted to minimize the cross-entropy of classification relative to the tested size of each patch (pixel-sized patches for example might lead to too high a training loss). Of course it might allow us to do a scattershot 'at a glance' type lookup of potential image contents, even if you get multiple categories for a single pixel, it greatly narrows the total span of categories you need to do subsequent searches for.

In other news I'm starting a new ML blackbook for various ideas. Old one is mostly outdated now, and I think I scanned it (and since buried it somewhere amongst my ten thousand other files like a digital hoarder) and lost it.

I have some other 'low-hanging fruit' type ideas for improving existing and emerging models but I'll save those for another time.