My Dunning-Kruger is so severe, I actually believe I’m smart enough to have Impostor Syndrome. 🤷‍♂️🤦‍♂️ Making the donuts at @panopticdotcom.

Butler, NJ, US
Joined December 2006
2
5
New "We Didn't Start the Fire" lyrics just dropped 🎶
fucking black hawk flying over my house, russian decoy drone found 90km away from where i live, ukrainian woman murdered on the bus, charlie kirk shot and dead, israel bombing doha, genocide going on in palestine, russian invasion of ukraine.
1
1
106
Replying to @HSVSphere
The idealists think good technology wins on merit. It does not.
4
10
2
229
11,900
Everyone dunking on "omg what a fool, boo tariffs, look at the markets run red" but first thing you learn: buy low, sell high. When markets are up so high, how do you do that? 1/
1
1
81
When you have the power to single-handedly drop prices 10% or more quickly, you go and do that. Then, buy low, and walk back whatever you did. Easy money. And, when you also control the SEC/DOJ, you can manipulate the market all you want. Follow the money.
1
63
I'm so glad key influencers have declared vibe coding "good" because it validates my argument that people who claimed "this uncommented, hard to understand code, all crammed in one long file, is bad" were focusing on the wrong things all along.
1
81
!!! ⌨️🥺
Guess which Blackberry patent expires on Saturday....
1
90
Just like there will always be people that know how to spell, the rest will just depend on spellcheck, and end up ducking and not know the difference between their/there/they're, and that's the future.
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
71
This is why I name my branches dev, qa, uat, stg and live. If you use more than 4 letters, you're wasting time.
Now that wokeness is over, can we go back to master being the default branch name?
1
90
And just like "cloud", someone will start marketing "agentless AI" to make it sound like they have something different. inb4 AaaS (Agents as a Service) is a marketing buzzword...
OpenAI appear to have settled on their definition of the term "agents": "AI agents are AI systems that can do work for you independently. You give them a task and they go off and do it." - @sama in the Operator intro video Does that definition work for you?
40%Yes, I used that already
12%Yes I'll start using that
19%No, means something else
29%Just see the results
1,313 votes • Final results
52
s/why their work is meaningful/how much money their work made and give them a generous cut of it/ s/be more meaningful/make even more money/ Sure, software engineers aren't only motivated by money ... but, it's definitely the easiest way.
The easiest way to motivate software engineers is to communicate why their work is meaningful and/or how it could be more meaningful.
1
67
This is demotivating not because it is true, but because it means you are working for someone who is wrong and likely incompetent. There's one thing to think your colleagues might be fools, but it's truly demotivating when they remove all doubt.
The easiest way to demotivate software engineers is to tell them their work is mostly fungible or going to be short lived.
40
Tell that to all the people who were still alive who were called back before Y2K to rewrite the COBOL they wrote.
Replying to @rakyll
All work is short lived
1
160
Giving me real Dr. Dobbs Journal Gimpel Software PC-Lint ad nostalgia. I loved those ads. Some months, only read the zine just for the month's ad.
If you know why x > y, you're a good C developer!
72
DTSTTCPW, in a nutshell. No, you don't need an ORM.
I am both surprised and not surprised the US Postal Service website's banner alerts are simply someone commenting and un-commenting the html. Random indents and all.
1
1
89
Serendipitous learning is undervalued. You can learn so much going down the right rabbit holes.
Replying to @zack_overflow
Interestingly, some devs actually valued the discovery aspect of finding information they needed to know Essentially, no go-to-def means you need to find it yourself, and the exploration can reveal important/interesting details in the docs or source:
44
Remember: If the product is free, the YOU are the product. These tools need access to more code to train on, esp. if they are forced to remove unlicensed work from their training corpus. Are you even authorized to give them rights to use your employers code? Better check.
Copilot is now free! Probably the biggest news for developers in 2024.
1
56
This is what happens when Perl programmers wind up in a C++ job and then create a new language that mashes them both up.
Rust syntax is disgusting, I’m sorry but wtf is this
1
1
161
AI won't be truly intelligent until you can tell it "arrange by penis" and it replies "you can't arrange them by penis." youtube.com/watch?v=uRGljemf…
things get pretty interesting when you give the AI access to your file system
53
> [...] most datacenter applications want to send individual messages to potentially thousands of recipients. Those applications should use multicast, then. Those who ignore history are doomed to repeat it, poorly.
What would it mean to replace TCP? We've been using TCP for almost 50 years, since it was designed for ARPANET in the late 70's. Back then, network links struggled to send kilobits per second, now, they send gigabits per second--literally a million times more flow. Through all that time, TCP has kept up, providing reliable connectivity across the entire world. I really like this paper because it convincingly argues that despite all of TCP's successes, there's one area where it no longer works well: the modern datacenter. TCP was designed for wide area networks where hosts are miles apart and take seconds to communicate, but it's being used inside datacenters where hosts are next to each other and want to communicate in microseconds. To make that more specific, one issue with TCP is that it's designed to stream bytes through long-lived connections, whereas most datacenter applications want to send individual messages to potentially thousands of recipients. Implementing these operations on TCP leads to high overhead, as servers need to find memory for thousands of mostly-idle connections and must use costly protocols to parse streamed data into discrete messages. At scale, these problems require elaborate workarounds, such as deploying fleets of proxy servers to manage connections and dedicating cores for message handling. As an alternative, the paper proposes a new protocol called Homa, a message-based, connectionless transport layer for the data center. It's practical because it requires no new hardware and can be implemented as the transport layer for common RPC frameworks like gRPC. John has been working on Homa for many years, so it's extremely interesting and there's a lot more about it in other papers. What's the takeaway? Software has to evolve with hardware, and especially the context in which the hardware is used. TCP has been incredibly resilient for our normal Internet use, but connecting datacenter applications is a radically different problem that needs new technologies!
42