I'll admit that it's a little shoehorned in at the end :)
... but you know how editors are with writing the headline for clicks against the wishes of the journalist writing the article. You'll always see Journos sayign stuff like "don't blame me, that's my editor, I don't write the headlines"
I did toy with the idea of going with something like:
`Prompt Engineering is a wrapper around Attention.`
I can't republish anything that happens in a production/proprietary environment.
One of the things that I think is pretty great about being able to share these particular prompts is that you can run this on one of your own repos to see how it turns out.
ACTUALLY!! Hold on. A couple weekends ago I spent some time doing some underlying research with huggingface/transformers and I have it on a branch.
I appreciate that this is long. You may have hopped on with your morning coffee, hoping to take in a few memes and hot takes before getting to work for the day. Then this monstrosity shows up.
Based on my beta readers' feedback of "too long" I've added things like a 2 minute summary at the top to try and let you know what you're in for, and a table of contents (sadly not working. it will when I republish on github) to hopefully let you skip to the bit that you're interested in.
You definitely don't have to read it all, especially as a front-line engineer but the leadership justification part is in there based on the pushback and "pressure testing" I've seen from people up and down the org chart.
As much as this IMO is an incremental shift in direction that should fit within a "normal" modified agile/scrum workflow, If you think about the cost of hopping onto a new workflow there's risk.
We're talking about an expensive bet here. Realistically your teams aren't going to know that this works for a quarter to a half. A long term shift in the way you do things takes time to sink in.
Costing out that bet on a team of 4 engineers an EM, PM and, Designer we're looking at say $20k/week/employee or between $2 and $4 million. You can fiddle the numbers around here since they're likely not going to get _nothing_ done but you get the idea.
Scale that to an organization and ... well it's a lot.
If leadership is going to allow you to adopt a whole new way of doing things, you're going to need a lot more rigour than just "trust me bro"
The size of the article in that respect is a feature, not a bug.
Leaders can hop off the train when they're satisfied that it's worth a try. The engineers can jump to the "just tell me what we're doing" bit.
I hope you read it all, have a robust discussion and really pressure test the ideas so I can refine the process. I'll take your accolades and your angry criticisms with equal excitement
(please be nice though, I really did try hard here).
As much as I appreciate the sentiment, it is starting to look like bad things happen to OTHER people when the extremely powerful get disconnected from reality. I feel like Marc Andreessen falls into the "bad things will happen to other people" category here.
As Keynes said, "In the long run, we are all dead".
Maybe Andreesen will manage to live long enough for life-extension stuff to really get going. But when you're a billionaire, there's a good chance you're gonna die before your stupid choices can do enough damage.
the mouseover tile expansion is way too big. if I like to hover my mouse over the thing I'm looking at as I scan the display, the expansion is so big that I can't see the next item to the right making the list painfully un-scannable for a certain UX.
don't zoom the tile. feel free to color-swap, display condition, whatever just don't mess with the size ESPECIALLY not a 200% zoom.
also - super buggy if I mess with the filters a little. looks like you're somehow registering "discounted item price" as "%savings" and reporting a $935.55 item as "93600% savings"
Thanks a lot for taking the time to report these points:
* I obviously cannot argue with you on the UI, it really needs some serious overhaul. I've been struggling because the images all come from Amazon listings and have no standard "graphical identity" so it always ends up looking like a giant bazaar of random items. The blowout tile was a test to make things pop but I agree it's not good.
* Noted on the zoom looking awful I'll change that too :-)
* Yeah every once in a while some items will register the wrong price so you get these funky percentages. For example I saw a pc registering at $1 for the average price so the computer discount is nonsense. The filters and sorting make those outliers raise to the surface. This should be easy to fix though with an automation to detect incoherent values and recapture prices. The number of occurrences seems to be low overall (i.e. 50ish out of 120k) but it looks amateurish if you sort on price/discount
I'd also add that the optimizing the timeline for engagement (including negative engagement & arguments) wound up making me quit the platform entirely. There were no positive interactions, just deeply unpleasant arguments with people spewing vitriol.
Like ... why would I stay on a platform that was so deeply unpleasant to interact with all the time?
reply