Weird that all of the discourse on this thread revolves around facts and figures about these various startups, and nothing to do about the inherent insanity of the “mundane thing as a service”, which is remains in my eyes as a bubble mid-pop
Not to be an egghead/navel gazer about it, but I’ve grown skeptical of “innovation” as an end in itself: was Facebook innovative, or was it just another small iterative improvement on an existing form? Same with Google and search. My gut tells me companies should focus on more concrete measures of success rather than the abstract “innovation”.
It’s probably not semantically wrong to say that these two cases really were/are innovative, but even so, was that really the cause of their success? And is it replicable as a methodology? My gut tells me that a lot of what gets labeled as a massive innovation is really just a market inevitability, and someone got to the right idea first, either by luck or having a single clever differentiating idea.
But it’s not changing their livelihood. Msft just gives them the same deal. In a lot of ways, it’s similar to the telepoll - people can just say whatever they want, there won’t be big material consequences
Doctors are somewhat unique in the fact that they can live and practice in a wide variety of places. Anywhere there’s a hospital, you will have them. Where they will choose to work/for how much is a much wider decision matrix than it is for many white collar organizations which would have no reason to exist (physically) in a small town of 20,000 people. Such a town would require at least a single hospital employing many medical professionals, depending on how far it is from other forms of medical care.
Mark Fisher had an interesting comment on this, and this is more in the context of public/privatization but the point is relevant, calling it “market stalinism”. In Stalin era Russia, bureaucrats spent enormous amounts of time compiling reports and window dressing for their project (essentially advertising its success back to the Party before it was complete). As a result, projects would be well known but mismanaged, slow, broken, etc. but the glory of USSR would definitely be upheld. I’m always reminded of this when executives get on stage and do the Steve Jobs thing about how great we are for working nights and weekends for the “mission” while only they have a vested interest in us making a profit.
Right, but obviously you can see the unknown unknown problem or Dunning-Krueger popping up when software developers who don’t know everything (unlike me, of course) start using it extensively without knowing the ways in which their prompt engineering is flawed.
Agreed. A good example of this (in Python) is to explore solutions to problems that use decorators/a more functional style vs. solutions that use classes and a more OO style. Usually both are totally good. The code GPT generates can be totally functional, yet inappropriate in context. Another good one is asking it to translate code from JS to a language without real “interfaces” and see if it uses a base class pattern or not. I review junior devs and lecture them extensively about stuff like this. (Yes, I am fun at parties.)
The U.S. is presently grappling with its own democracy-related issues, rise in authoritarianism, and declining quality of life, as well as a lot of “de globalization” problems originating in its poor attempts at playing hardball with China and other powers (Russia) that it just doesn't like.
In my opinion, the most grim and dystopian elements of China’s system (Social Credit, censorship, mass surveillance, extreme bureaucracy) are either already present in the U.S., but with a decentralized/capitalist bent rather than a centralized/communist one.
To play devil’s advocate: have you never reviewed a PR and said “yeah this works and completes the ticket but you REALLY shouldn’t be doing it like this”?
There’s a pretty big gulf between functional and correct in a world where optimization matters. This is not always the case (strictly, i.e. algorithmically speaking), but you should be optimizing for things like reusability, testability, etc. which GPT-generated code can miss.
Imo if you’re “doing coding right”, at a certain point, you should be thinking very little about what you’re doing. This can manifest as having easily copypastable design patterns, lots of tests, inversion of control, etc.
Even among senior developers it can be shockingly rare for people to actually understand and implement something like DI/IoC in a correct and useful way.
If you ask it to start explaining/extending/implementing classes/interfaces, start asking it to reason about leaky abstractions or when it’s appropriate to use Mixins, it frequently generates code that either doesn’t work or just misses the point. It’s especially hard because some of these problems actually are not tractable in the language you’re taking about due to their features, or due to limitations of your code/your knowledge it may suggest you do something slightly worse because you haven’t prompted it on factories or other design patterns that you might want to use.
Yeah the democratization era is already well underway: arguably over, even, in that it’s no longer possible to become a “big star” from throwing together some shit in fruity loops and hitting it big. Every once in a while, a bedroom musician with a little talent will break through, but more and more music is either relegated to “good taste but unknown” or “weird conglomerate of ghostwritten tunes operated by a Hollywood finance group”. AI might as well already exist. Half of what you find on soundcloud is low effort/low skill/low expression anyway. Which I think is great.
fwiw there’s a ton of upcoming (and heavily debated/not yet implemented/understood) regulations around EV chargers that will absolutely nuke some of the shittier app-centric/not really functional charging systems that people rely on today. I’m not a Tesla fan per se, but it is remarkable how much better life is for Tesla owners vs. the huddled masses roaming the frozen wastes looking for a charger that can actually dispense a charge or provide users with some useful error feedback.
Fwiw locking people up actually does make some addicts lives easier. The life of a hard drug addict is filled with the constant strife of trying to procure money/drugs, all while extremely fucked up, and the. navigating the dangerous and shitty situations that puts them in. My aunt went to jail for 5 years for meth and she said it was the first time she felt she had enough piece to actually understand and work on sobriety since she started using 10 years prior. She’s been close to 20 years clean now. Granted, not everyone has it so nice, but she lost her kids, became a felon, and suffered a lot because of it. But then again, meth addiction begets a lot of suffering as well.
It’s already currently true that many mid-tier shops employ an army of low-knowledge practitioners who will always “solve” whatever problem you give them in record time by abusing the tools you give them, reinventing wheels, punching semi-truck-sized holes in otherwise functional abstractions, barely testing anything. All is well until scaling problems or distributing desyncronization bugs or severe data-loss/replication happens in production.
Once you’ve seen this happen, and especially when you see it cause outages that costs thousands of dollars, you understand why it’s worth it to “pay more than you need”. When the rubber hits the road, having an army of automata who “get things done” is functionally not the same as having skilled developers who own their craft in the long term.
If I were starting a company today I’d probably be fine taking on some tech debt to get v1 out the door and then worry about investing in a dev team who can scale/rewrite it into the version that can scale to whatever level I need. But in no way would I want to ever again watch a junior dev who doesn’t understand how to read logs trying to implement a caching + crontask solution to reduce app load times from 30s to 15s on a backend query against a table that holds 10k records, because their code retries 500 times because they don’t understand timeouts, indices, or ORM induced n+1 issues.
In the long term, automaton armies will always curse you to the problems of local min/maxima problems unless they are backed up by someone with enough global vision to get them out of that hole.
I frequently run into this, and it’s quite maddening. When you’re working on a toy problem where generating functioning code is giving you a headache - either because it’s complex or because the programming language is foreign or crass - no problem. When you’re trying to extend an assemblage of 10 mixins in a highly declarative framework that many large-scale API contracts rely on to be correct, the problem is always going to boil down to how well the programmer understands the existing tools/context that they’re working with.
To me, a lot of this boils down to the old truism that “code is easier to write than maintain or extend”. Companies who dole out shiny star stickers for producing masses of untested, unmaintainable code will always reap their rewards, whether they’re relying on middling engineers and contractors alone, or with novices supercharged with ChatGPT.
If it’s the difference between them getting your $50 per month network subscription you bet they will disable your headlights if given the legal leeway to do so.
All functions that a consumer could reasonably expect a car to perform without an internet connection must be operable without an internet connection. All features related to driving the vehicle, including but not limited to steering, braking, using traffic signals, starting/stopping, toggling locks or lights, opening or closing doors or compartments, and other standard features of a non-connected car must not require a network connection.
I think it’s probably OK if automakers want to onboard and support network features like maps or outsourced computation/diagnostics, but bricking everything because the company wants to collect user data and they want to opt out is obviously unethical. It’s important for these regulations to include the fact that they cannot be worked around or stapled on by the auto company post-hoc via firmware upgrades or the like.
I tend to agree that both practically and legally, when you’re dealing with a fraud as big as this one, that successfully prosecuting the Big Man At The Top is the best strategy.
People might scoff at mid-level fraud “doers” getting a slap on the wrist deal for their testimony, but the alternative where you go full scorched earth and fail to get a conviction would be terrible for the defrauded and the system beyond them. Capitalists (i.e. the owners of capital, the investor class, in this case) love a system where you don’t really have to quadruple check that a financial document is valid. Who can blame them? I certainly wouldn’t want to live in a world where something like FDIC doesn’t exist or where fraudsters can easily rip people off with low chances of jail time if/when the operation goes belly up.
Communism was/still is (depends on how you feel about China) a system that billions of people lived through. I think there’s been a great amnesia about this in younger people who’ve never met someone who interacted with Really Existing Socialism
Yeah, but there are obvious differences. A lot of people watched TV with their families or friends, although this is likely less frequent now with streaming. The amount of interactivity, portability, and solitude that the iPad affords its users with is undoubtedly bad for developing minds, at least in large doses.
Essentially, yes. Between a few times per day and a few times per month I would like to create branches to store the “backup” copies of table-stored confs to compare against logs/events and see if configuration changes caused certain conditions to arise. In the case of a configuration error, rolling back to a prior state would be awesome.
Hi Neon CEO. I’m curious if users have deployed Neon as a (selective) audit-logging tool for configuration which might need to be edited or rolled back from time to time. Think configuration-driven/event-driven systems where an update to a single database-stored config can cause other downstream applications to start behaving mysteriously. Going to scour the website some more!
I dunno, I’ve never been head-over-heels passionate about writing code, though I’ve been at it for 15 years now and I still get quite a kick out of it in the right circumstances. I tend to gravitate towards bug fixing, integrating work others have done or improving it slightly, testing, deployment/monitoring automation, etc.
I’ve also 100% had lulls where I just hate it and cannot be bothered to do my job or do it well, I think it ebbs and flows with what you’re being asked or required to do.
I don’t think that’s really ADHD. It’s a motivation/interest issue (and maybe it’s not really an issue) and it can be fixed by factors like choosing to pursue fewer things, focusing on your day job, or prioritizing certain things over others. It might be ADHD if they’re not functional at any of it, but it certainly doesn’t sound like it, and from my experience with people with ADHD, many are actually dysfunctional to the point of needing medication to get anything meaningfully done.
I would absolutely kill for a job that’s 2 days per week in office with everyone on my team in one room. It’s really amazing but it’s hard to find anything but places that do “zoom from open office”, which is nonsense.
Sadly people are and do. I got a text earlier saying that my bank (and they, by random chance or something more nefarious) named the bank in the communication AND they got lucky because I was just traveling nearby, AND before leaving home I had just opened a piece of mail that informed me that some piece of shit company that I got concert tickets through just had a data breach AND that a bunch of my stuff, including my credit card info, had been leaked. So I definitely had that moment of “oh shit, I should probably look into this” - and if I didn’t have my bank app on my phone I may have just fallen into the process. Scammers operate a simple sales funnel - they cast a wide net and hope they get someone in the same headspace as me but slightly more dumb. The only people who fall for this are, sadly, really dumb, but also really unlucky because they might not have fallen for it if the other coincidences didn’t align.
True - but there’s been a ton of stylistic and technological advancement of the basic 88 key device over the years - ranging from the development of key synths to (neo)classical reimaginings of how to use the hammers and strings to do something novel, not to mention AFX going nuts with little servos and player pianos, plus the development of jazz, ragtime, etc.
One practice I’ve gotten into with juniors is checking out their PR, branching it, PR-ing to their PR, and then having the discussion about what I want them to change there. This is good because it lets us isolate certain issues (and truly resolve them) rather than comments getting pulverized by lots of little commits, only to have someone else come in and say “LGTM!” and merge it with unresolved issues.
Another obvious way to fix OP’s issue is to request developers to break their PR’s down into smaller chunks, pair on the parts that need help, and slowly merge things from there. YMMV of course - but any 1000 line PR is going to be a headache regardless of what methodology you’re using to review code.