HN user
Very cool. I developed something in a similar vein as a way to teach myself web programming 15 years ago or so. Https://dirtywalls.com is location-based message boards. You can create or join ones close to your location. Reminds me that I need to try to tell people about it since it’s mostly just me checking in to my local bars and shops.
Curious if you got a response as I'd love a parameterized version of this design. I have the original Overwork+Scott version on my bedside, but I'd much prefer a storage space than a charger there.
A relatively small ongoing investment in a phone with which they earned billions of dollars in profit. Doesn't necessarily require new feature updates, but security updates should be available for a far more significant length of time than the single-digit years the have self-regulated themselves. As an alternative, perhaps these companies should be held responsible for the e-waste of their prematurely expired hardware...
Not a SDE, but I’m actually doing more of that than ever before. 2 years ago, was given the choice to interview for new internal roles or walk away with a severance package after almost two decades in QE and PM for a large, design tools company. Having witnessed the transition from developer-led (“hey, I just invented this cool thing. Let’s give it to users and see if there’s something there?”) to product-led (“hey, let’s keep making our awesome products better for customers!”) to ultimately marketing-led (“hey, McKinsey taught us this method to circle the wagons and protect our brand to the detriment of paying customers!”), I chose to walk away.
I had a bit of cushion, sold some early-purchased crypto and spent most of my savings, to build my skills in launching a startup and writing code. I interviewed with plenty of companies, but felt the same McKinsey cringe from the big FAANG’s, couldn’t believe the skeeviness of the crypto startups, and went pretty far but not far enough with some of the smaller but more interesting companies. My startup got some interest but ultimately, did not hit hard enough to break gravity.
But, it kept me on the radar for some old contacts and when they wanted to explore the areas I had expertise in, reached out. I did some contract work, ultimately building and launching a major new product on their platform which led to getting brought on full time at the beginning of the year.
I love what I’m doing now. I get to play PM and R&D and Developer roles building things now. The things I build directly impact real people and I can experience the results within days, and I have flexibility to decide what I think is important or interesting.
Oo! An opp to share a project I built! Please check out ottomusic.ai or the original demo I built at demo.ottmomusic.ai. We were one of the first AI-powered, prompt-based music recommendation service, and I still think far better than Spotify or Amazon’s recent drops.
(For users of the demo site, you can get a Spotify link to the generated playlist in the console. Shhh.)
How odd. My 3rd click, it brought me to news.ycombinator.com - exactly what brought me there in the first place and open in another tab. I was very confused as I assumed the tab had closed but no, it just appeared to be luck of the surprise.
One day mid-1995 (just looked it up and it would have been May 20, 1995) Sonic Youth was playing with REM at the Gorge in Washington. He and the drummer had a side project that came to the little record store in Ellensburg (Rodeo Records, owned by Mark Pickerel of Screaming Trees fame) for a secret daytime gig. I lived half a block away and was tipped off by a friend who worked there. Went and drank with them all after the gig. Class acts all the way
Any eye toward allowing users to train their own voices? The only reason I’m using Elevenlabs is because I can train a suite of voices on my (legit, legal) archival content. It’s not a perfect replication of the original voice, usually, but for my purposes, this isn’t a requirement. What it does get is the artifacts, recording environment, and a large swath of the prosony and other voice elements that make it sound real and not AI
Recently my son and I were at the mall and decided to grab dinner at a popular ramen spot. They had replaced the hostess with an iPad and a waitlist app. Ugh, I thought, but fine. I can type my name and the number 2 on an app just as easily as letting someone write it down.
Except it wasn't as easy. The iPad first demanded that I make an account with the waitlist app service. I could do it on their iPad but it encouraged me to download an app on my own device instead. There was no “no thank you, just get me a table please” option.
We noped out of there pretty quick and even avoided it the next visit to the mall. I don’t know what kind of money they’re saving by switching to this system, but they’ve definitely lost a $60/visit customer who would probably stop in 3-4x per year.
And I just realized you acknowledged it at the bottom of the page! ;)
I was fortunate to help bring similar tech to life as a PM for Adobe Audition and Premiere Pro in a feature dubbed Remix. Since we were designing to help music fit the duration of a scene or video project, our goals resulted in a slightly different experience from those of music fans wanting to recut a song for general purpose listening, but it worked like magic for most tracks. Still one of my favorite projects that I worked on.
I relied upon ChatGPT to build this as my development experience is pretty light. What I found was it's not perfect, but because I know enough Python to troubleshoot, I was able to either fix issues myself or understand the right way to frame the next prompt to improve or resolve any problems I ran into.
I started a new chat and explained that we would be working together to build a web app that would take a user's input and use the chatGPT API generate a unique music playlist, taking into account every mood, activity, emotion, attitude, era, and artist that the user mentioned. The results would be displayed in a table with the album thumbnail, artist and song information, and an audio preview. The user should also be able to open the full playlist in Spotify.
It recommended we create a Flask app in Python and immediately generated some code and instructions. It took a bit of finessing to get the right libraries installed and setup the environment variables correctly, but I had something working within about an hour. It was ugly, it was slow, but it was functional.
After several queries, it's recommendations no longer seemed to be talking about the same code we'd written. It would reference functions that didn't exist or weren't named the same. For awhile, I could reset it by simply saying, "Here is the current version of the code" and paste the source in, but eventually, it got too large. I refactored both the python file and the html templates so that they were more manageable sizes, and that helped. I found I could also simply paste a function or section of code I was having trouble with and usually get usable feedback.
I had avoided using frameworks like Bootstrap before because there seemed to be a learning curve that I just didn't have time or need to dive into. However, because it designed the app around those frameworks, I had the luxury of just seeing the results and how they were achieved. I won't pretend I'm an expert, but I do feel far more confident if I needed to start from scratch.
By the evening, around 6 hours after I'd started, I had an incredibly usable web app. It was clear enough that I could continue to iterate on my own, occasionally asking chatGPT for assistance when I wanted to make a larger change. Over the next few days, I worked with some partners to fix up the design and hosted it on a server that wasn't my dev computer.
About a week before we launched, I ran into a bug. Two queries ran very close together and the first user got the second users results. Oh no! I realized I'd need some sort of user session management on the server, but didn't really know where to begin. I pasted some of the API functions to chatGPT and explained what happened and what I thought needed to be done. Within 15 minutes, I now had server-side session management. But now I was running into openAI rate limits! It took a little longer to get that all working, but within an hour or two, I now had a working queueing system.
I was then able to very quickly add little bits and pieces: a button to use speech recognition to submit your request instead of typing, the ability to share your playlist via SMS or notes, and an option to extend your playlist if you wanted more songs.
I feel like I've learned more in the past 3 weeks than I have over most of my technical career - I was certainly able to achieve success much more quickly which kept me motivated through the more difficult or confusing periods. The response today has been fantastic, and I'm really proud of what I created.
If you haven’t made your own butter at home, you’re missing out. It literally takes 10 minutes and the blender does most of that work.
Pour a cup or two of heavy cream in the blender. Add a tspn or so of salt, and blend it on med. it’ll turn to whipped cream and then chunk into butter. Let it run for a minute or two more as the liquid co to yes to separate from the butter.
Scoop the butter into a non fuzzy towel or a few layers of cheesecloth and squeeze the remaining liquid out. Spread on hot bread or grab a spoon.
A short seller borrows shares from a shareholder - a bank, institutional investment group, or an individual who makes them available in return for interest payments - and sells them on the market at the current price. They hope to return the shares by buying them back at a lower price later and pocketing the difference.
I worked at one of the bigger companies (not a FAANG, but right near there) for almost 19 years before leaving last year. When I joined, it was as amazing a company and culture as I’d imagined, having used their products for years and being a big fan. It was very engineering-driven, with brilliant people being given a loose rein to explore and invent. Cool inventions would find their way into one of the products. Incredible ones would become new products themselves.
Several years later, after some meaty mergers and exec shake-ups, it changed into a product-driven culture. It was less about boldly exploring and more about shoring up the product line and making sure it was as all-around essential as could be.
Then, the MBAs swarmed in. Average tenure went from double digits to mid-singles. Another round of exec shakeups and the new order was clearly marketing: protect the franchise at all costs. VPs who had shone a little early on were now erratic and damaging - to products and morale - with little attention from corporate. It was not about creating amazing things. It became about circling the wagons and squeezing every fraction of a penny everywhere possible.
Working on a more niche line of products, it was an almost Brazil-like routine of pointless effort being praised, ignored, then quickly forgotten. When the trends presented became apparent, then it became a blame game of who was at fault, rolling downhill, of course.
It wasn’t until I was actually on my way out I could recognize how toxic it all was. Like a cancer that grows so slowly, you don’t really notice that small bump has become a bowling ball sized tumor on your neck.
Not sure what advice I have other than to trust your gut, and put aside any loyalty or connections you have to the things you helped create because they’re not yours and will be handed over to the next person without any fanfare. You are not the company you work for. Disconnect the feelings of identity and cut the umbilical as soon as you feel it attaching itself.
I was 16 or 17 and had a guest account on the local universities Ultrix server. I discovered that all of /dev/tty* was world-readable UNTIL someone had successfully signed in.
So a “cat /dev/tty* > passwords.txt” and waiting an hour collected the credentials of everyone logging in to the server.
At some point, I had logged into the account of one of the sysadmins who msgd me and let me know he’d changed his password and this would be my last time on his account. (If he only knew.). He offered me my own account if I told him who I was, and he might have meant it, but I didn’t bite. About a month later, a patch fixed the issue and that was that.
If you used and miss Spotify Run, Muze is a lot like that only WAY better. Not only is the music by real artists that you know and love, as well as new discoveries, but the responsive aspect is unlike anything else that’s come before.
Spotify Run measured your initial pace and then “locked in” and only played tracks at that tempo. With Muze, we adapt to your pace, so the faster you run, the faster the songs play to match. Matching music tempo to a runners steps per minute proved to be a difficult challenge, but it makes such a difference to maintain pace and push through tough spots.
If you try it out, please let me know your thoughts and feedback. I’d love to know whether we’re hitting or missing the mark.
I couldn't stand the Cal Worthington commercials, but I'll remember them fondly because a sweet, elderly neighbor with dementia misunderstood the repeated phrase at the end of the song, and would loudly sing along with some rather improper words that were VERY funny to a kid.
In addition to our friends in the diamond business, we also had Jack Roberts Appliance who was less offensive than the others.
Lots of possibilities, for sure. I liked the idea of a public layer as well as private or protected layers for every space. Patrons of a restaurant could chat on the public board for that location, while employees might bitch anonymously about management or warn each other of problematic customers on the private space. Or residents of an apartment building could complain about noisy neighbors or neglecting landlords without it being completely public.
My concept had a sort of anarchic bent at the time, so anonymity and fighting the man was the guiding principles in theory, but as I was only just learning JavaScript and PHP, ideals were going to be a later development and design phase than I ever achieved ;)
About 10 years ago, I taught myself web programming by writing https://dirtywalls.com which is still up and which occasionally I still see used by a handful of the folks I shared it with back then. Same basic idea: pseudonymous message boards attached to real-world locations, where you must be within a certain proximity to read or post.
The concept at the time was "4chan meets foursquare." and the inspiration was rediscovering some graffiti in a bar bathroom that years before had particularly tickled some friends and I. For awhile, I added new locations as I traveled extensively for work, so there are hidden little message boards around NYC, SF, China, Seattle, Europe...
Venkatesh Rao also put out a book collecting his newsletter, "Be Slightly Evil" with a lot of explorations of this topic like, "What's the right length of time to hold a door open for someone?"
Another book of his is Tempo, which is more about understanding and manipulating the tempo of different interactions in business.
But definitely start with the Gervais Principle. It's wonderful.
Shame that the mobile friendly library can’t demo on my iPhone because it thinks the resolution width is too low
https://www.modeltalker.org/vrec/ is a project for "voice banking" that might be able to help. It's not perfect yet.
Let me add that it might be better if the chatbot WERE acting as an agent, referring to its owner by name, instead of pretending to BE them. For example, when asked "What are your areas of expertise" it might be better if the chatbot responded with something like, "diggum has 15 years experiencing designing creative tools for designers..." Right now, it's a bit split personality responding as the bot for some things and as the owner for others.
I love the idea. At this point, I wouldn't put it on my profile, but can see the value in doing so once it's progressed.
Right now, much of the interaction is with the bot itself, and not the person it's representing. As a questioner, I don't care how the bot is doing - I would expect the answer to come on behalf of the bot owner. The humor around "asking Alexa in a few hours" was unexpected, and the links provided didn't seem related to my original question.
Also, please support submitting the question when hitting Return, rather than forcing a click on the submit button every time.
Again, I think natural language chatbots can be a big deal, acting as interactive agents that represent their owner for all sorts of situations. But until that interaction is smoother and slightly more believable - not "out of the uncanny valley" perfect or passing a Turing test, but closer to an actual dynamic conversation - it's tough and provides less value, more slowly, than simply scanning profile text.
The messengers are usually flawed, but the message is good. Same with so many other things which can positively enhance our perspective: comedy, art, music. The messengers are one of us, and while they’ve often glimpsed a possibility most of us haven’t seen, they’re still bound by the same limitations of humanity the rest of us experience. Their message can be guideposts toward a better existence, even if their examples fall short. Finger pointing at the moon, and all that.
Definitely in my top ten list, too. Like Machiavelli, but its D&D alignment would be Neutral Good.
Shortly after a big upgrade to the application I worked on, we started to get bug reports on our forum about a pretty ugly bug where a modal window would get “stuck” after clicking OK. The window would no longer respond to users but the application would act as if it were still up, and the only way out was to kill the process and lose all the unsaved work.
Understandably, the complaints got louder and angrier as no one on our engineering or prerelease teams were able to reproduce it, but easily half the bug reports were all about this issue and identical. After 2 weeks of trying every configuration of video card, OS, and any other alchemical recipe we could think of, my partner on the test team stumbled upon the license crack online while searching for anyone else having this problem but not having reported it to us directly.
As a brief respite from this ongoing madness, he decided to install the crack on a clean machine and aee how it worked so we could report it. Lo and behold, the problem reproduced on the first try. This whole time, indignant, angry users who were actively threatening our team and vowing to never use another product from our company had been bested by a poorly written crack.
Before exposing the secret, I asked dozens of the most vocal and vicious reporters to please email me a particular log file, and give me contact information where we might get in touch with them as we were close to cracking the case. (Pun intended, and this was before we had very clear rules around personally identifiable information, but even with the policy at the time, these weren’t our customers so I felt less anguish about it.). I received all of it and more, invited them to share their experiences on the forum threads, and then tactfully but clearly explained exactly why they’d seen this problem. I offered to work one on one with anyone who felt my accusation was in error, but not one single affected user replied or followed up after that message.
Had an Amazon Restaurants deliver a pizza sideways in his shoulder bag. Opened it to show him the pizza was now a clump at one end of the box and that I was going to need him to try again. He called and seemed very confused as to why I wouldn’t accept the pizza in that state.
45 minutes later and had a flat replacement delivered with a refund and credit from Amazon. Other attempts from other services have typically arrived cold and late. Allowing me to see notifications as to when the food is ready, then wait in agony as nothing happens for 30 minutes, gives me a pretty good idea what I’ll be getting. So far, I’m not a fan.