> He was someone who played games and manipulated his way through his career. He could say the right things and navigate his way through office politics unscathed while causing damage to everything he touched.
Not the one you're replying to, but is the above quote not enough to explain why they were a mis-hire?
I think there's nuance missing from his take that people gloss over.
To me it's more like "don't over-invest on the early thing you're going to throw away". That implies that:
- You're making a thing, and making it quickly - optimizing for rapid market/customer feedback
- You're making it in such a way that you can replace it
- You have real plans and a sincere intention to actually replace it if you start to scale
I have no problem with those, but what I see at a bunch of startups is:
- Make a thing, don't worry about paining yourself into a corner because "we're moving fast"
- The thing hobbles along, the product gets some interest
- It's incredibly hard to maintain, replace or otherwise improve
- Original authors leave to other pastures because this isn't fun any more
- Other folks come along and have to work miracles to try to undo the gordian knots that were made
In other words: I like the idea of moving fast early on. Duct tape programming, love it. We also have to pay attention to the long term impacts of the decisions we're making. Some things are easy to fix/replace/undo, others are brutally hard.
Say you write a prototype in {Ruby/Python/Perl/??}, now you have mass adoption and you rewrite a slow part of it in {sexy_modern_language}. That's great.
Say instead that you decided to store all of your data in some bizarre format, hosted on IPFS, accessed via bittorrent, through an onion network, via ZModem, with CORBA - and you bake this in to such a degree that it's intermingled everywhere. Now you have a dilema:
- you have a business, making money
- you have increasing interest
- you have a system that's effectively impossible to change or reason about
- you can't really improve things to match the customer interest
- you're wedded to the design because things are so interwoven
- your only option is essentially a ground-up redesign and rewrite
Now - if the product is simple enough, rewrite away! (think early Twitter), but if it's not? Good luck to you.
I've seen this pattern more times than I haven't, and I think it deserves more nuanced thought and care when it comes to the economics at play.
Please show me one doctor who recommended taking a rock each day. LLMs have a different failure mode than professionals. People are aware that doctors or therapists may err, but I've already seen countless instances of people asking relationship advice from sycophant LLMs and thinking that the advice is “unbiased”.
Homeopathy is a good example. For an uneducated person it sounds convincing enough and yes, there are doctors prescribing homeopathic pills. I am still fascinated it still exists.
That’s actually a example of sth different. And as it’s basically a placebo it only harms people’s wallets (mostly). That cannot be said for random llm failure modes. And whether it can be prescribed by doctors depends very much on the country
Actually, swallowing a rock will almost certainly cause problems. Telling your state medical board that your doctor told you to take a rock will have a wildly different outcome than telling a judge that you swallowed one because ChatGPT told you to do so.
Unless the judge has you examined and found to be incompetent, they're most likely to just tell you that you're an idiot and throw out the case.
> the money still flows if you do a bad job so why do better?
I'll raise. The money flows because you do a bad job. Doing a good job is costly and takes time. The money cannot invest that much time and resources. Investing time and resources builds an ordinary business. The money is in for the casino effect, for the bangs. Pull the arm and see if it sticks. If yes, good. Keep pulling the arm. If not, continue with another slot machine.
I would argue that the money is short termism though. It just assumes short term returns are the correct choice because it lacks the technical understanding of the long term benefits of a good job.
In my experience many acquisitions set the peak of a given software product. The money then makes the argument that its "good enough" and flogs the horse until its dead and a younger more agile team of developers eventually build a new product that makes it irrelevant. The only explanation I have for why so many applications fail to adapt is because of a cultural issue between the software and the money, that always gets resolved by the money winning.
For example I would suggest that the vast majority of desktop apps, especially those made by SMEs, originally in MFC or something fail to make the transition to online services that they need today because of this conflict. The company ends up dying and the money never works out what it did wrong because its much harder to appreciate those long term effects than the short term ones that gave them more money at the start.
reply