jongjong

But how many new ones did it introduce?

Especially true now with AI coding. But I would push back on the idea that sciences or other engineering disciplines are not affected. Several of the people I met who were especially guilty of this in software sector had a PhD in mathematics and used to work in academic institutions.

I think intellectual integrity has been on the decline overall.

This article highlight a huge problem which permeates every aspect of society.

The entire education system is built around the assumption that speed = merit. Any time a person has to sit a test under time constraints, the most significant factor being measured is thinking speed. Not reliability, not creativity; just speed.

I have similar thoughts about 'short term thinking'; this is another religion which has quietly taken over nearly every aspect of the modern human experience.

Sure, but you can only solve a problem that you are aware of. My concern here is about the unknown unknowns. The task at hand is exceedingly complex and requires extreme precision, with many unknown variables... And the people solving the problems have conflicting interests as they are paid hourly.

Even if there are bonuses, all the bonuses would be given on the basis of solving known problems. There is literally zero monetary incentive to solve unknown problems; only owners care about that stuff. Employees don't care about that. They might point out some whilst not mentioning others.

I've been a software engineer for 15 years and my experience is that unknown unknowns in tech are like a secret piggy bank for engineers to tap into in times of need. They are rarely brought up; only in the event that an engineer feels the need to justify their value to the company/project. The accumulation of secrets gives engineers differentiable value within the project/company and makes them hard to replace... So they tend to keep at least a couple to themselves... They ration it; like retirement planning.

The individual incentive to conceal issues until a time of personal need (when layoffs are looming) outweighs the collective incentive to identify new issues to generate more billable hours of work for the team.

If you're the only person who knows the issue which prevents a multi-million dollar project from coming to fruition, that is valuable information and gives you a lot of leverage in terms of job security... Which is the biggest reward you can get in this line of work.

And it's not malicious; the people who know about those issues probably also solve more issues and contribute more solutions than anyone else. The concealment may be happening subconsciously; they won't admit to themselves that a specific kind of problem exists.

Like when I worked in crypto, all the engineers there believed that the project could scale and also be decentralized... When it couldn't, given the chosen approach. The belief is still there years later; though all the rational engineers left, they were promptly replaced by believers.

Makes me think. I can't be as confident there as I got lucky in that respect so my perception is distorted towards "it's easy". Based on what I observe from other people, I would think so as well.

I think relationships are heavily based on who you are and what you project and it's extremely nuanced. Some trivial physical or personality feature that you have can radically shape people's perception of you and there's often nothing you can do about that.

Sometimes a (perceived) flaw that you have can actually work in your favor as the other person may see it as a benefit. Relationships are as much about attracting the right people as they are about repelling the wrong people.

The framing that there is some unknown particle affecting the results is probably an understatement. Probably there are a LOT of unknown 'forces' affecting the results. Look at the size of the machine, the aging components, the number of components involved, the amount of software required, surely all of that is affecting the results.

Are humans even capable of building perfectly reliable systems?

My biggest mistake in life was thinking too big and aiming too high. Now I absolutely hate the saying "If you want to reach the moon, aim for the stars." This belief ruined me.

The people who were writing desperate emails and bending over to get a wage-slave job are much better off now.

Meanwhile I was trying to raise funding for big ideas from big investors; it was a complete waste of time.

Reality is "If you want to reach the moon, aim for the clouds." Then you have a better chance of actually achieving something, even if it's not your first preference.

I think that by the time AI can fully automate software engineering, it will be able to fully automate physical electrical work.

The challenge in both cases is all the invisible, hard-to-describe stuff that people are doing which business people take for granted.

Business people don't consider "minimizing complexity" to be something software engineers do, but it's literally the most important part of the job for any large complex project and it's almost impossible to measure. Even humans are pretty bad at it but at least there is a limit. With AI, there is no awareness of this and more complexity means more tokens to sell today and in the future to maintain so there is a perverse incentive and no guardrails.

Also, software engineers make a lot of decisions based on human factors and factors that are unique to the company/project.

For electricians, the part which AI struggles to automate is robotics and stuff like travelling to the job location, ensuring the right cables are available, identifying the right port to plug the cables into, drilling holes if necessary, climbing into the ceiling, pulling rolls of cables along the ceiling or wall with appropriate attachments, setting up the electrical boxes.

I think in the near future, all manual jobs will be replaced by a combination of specialist robots and generalist "human minders"; a minder will be transferrable across any industry and the entire job will be exception handling any unexpected issues which the engineers didn't yet train or program into the specialist robots.

I don't see the need for MCP for most use cases where it's currently used.

You can get better results with skills SKILL.md + linked .md files with curl commands inside... Just plain HTTP(S). I say this as someone who often prefers the stateful WebSocket protocol for data transport. HTTP stateless request-response model is a natural fit for LLMs.

HTTP is an excellent protocol for this and curl is an excellent, very popular command on which all major coding models are well trained. LLMs are insanely good with curl. Not to mention that it's pre-installed on all major operating systems. The curl command has become the protocol.

Then there is the fact that most tools require some documentation to go along with it anyway (to be used correctly). MCP mostly adds unnecessary work for a lot of cases.

It has its place in backend code environments I guess... But most of these use cases are thin applications trying to be a middleman between the user and the AI. I think ultimately, the AI will be at the front and tools at the back. Hence, curl.

Wedging a platform between the user and AI creates unnecessary constraints and reduces interoperability and integration opportunities.

These are all valid points but not strong enough to warrant banning open models IMO. This is a new reality we have to live with, with new dangers we cannot avoid. Best solution is improving the social system, lowering the incentive for people to do harm, restoring social trust, etc... This is the only way I see the US getting out of it. We really need a high trust society now. This is the only protection. We need neighbors looking after each other. There needs to be an incentive for people to do the right thing. Everyone should want to protect the system. Coercing people into not doing harm is never going to work long term IMO.

When people say "What will be left for us to work on once AI takes over", I say "Ourselves" - We have to stop looking at humans as commodities and strategic pieces and start paying attention to people as individuals, based on the content of their character. We need a society which is attuned to this, which has enough resources and time to pay attention to this. Now we are blind to people's character because it is masked by money, power and status; all of which currently have higher priority. This order of priority is determined by scarcity, which is largely artificial.

We have a dishonest system which tries to control people's behaviors through scarcity-based coercion instead of straight forward laws or simple incentives and clear explanations.

I guess in both cases, it's the result of the mind adapting to a strange environment. Once the strange environment becomes the new normal, the old (previously normal) environment seems strange.

Probably in the first few days in the submarine or space station, it feels strange but you don't pay so much attention to the feeling because you expect it; it's unfamiliar in every aspect so there is no internal dissonance. After coming back, the old environment feels different and that seems strange since you didn't expect it to feel different; it creates a dissonance between your body/senses/subconscious and your conscious mind. Your conscious mind hasn't changed but your senses and subconscious mind did. Going back to the old environment makes the change within you seem more obvious since you have a conscious frame of reference to compare it to.

This is probably why the conscious mind finds itself observing the work of the subconscious mind with skepticism. There is probably some kind of loss of trust since the conscious mind observes that things don't feel the same as last time despite knowing that they haven't actually changed.

I feel like democratizing wealth and retirement has become harmful.

Now we have a small number of extremely influential people who are essentially bribing and propagandizing the masses into voting them more money... Each win gives them even more money to buy even more votes at the next election.

Politicians change but the people behind them keep accumulating more wealth.

Democratization of wealth creates massive capital centralization which creates anti-competitive dynamics; the monetary playing field is itself inherently anti-competitive by design.

I often give the example of a skilled coffee shop owner having to compete with a coffee shop 'pet project' of a multimillionaire passive investor who is running their shop at a loss... The person without passive income cannot win no matter how much better they are.

Now extend this metaphor to every aspect of the economy. There will be two kinds of businesses remaining; monopolistic megacorps receiving government handouts and small businesses which would all be some millionaire shareholder's unprofitable pet project.

Also, the biggest shareholders of various companies are using their leverage to find anti-competitive synergies between companies in their portfolios and coercing the executives of certain companies into favoring specific suppliers and clients on that basis.

My position is that prediction markets are bad and harmful but so are all mainstream stock markets. They're all prediction markets. People collaborate to pump up the value of specific assets or contracts, leading to undesirable outcomes.

It would be dishonest to ban prediction markets without banning stock markets.

The best which can be done, under the current system, is to punish unfair information asymmetries, which is practically impossible to do reliably... And even if we could find a way to punish it reliably; what would happen is that most rich people would end up in jail. Information asymmetry is almost never fair and it's how people make big money these days.

I think the best solution is to create conditions to allow people to earn big money in other ways; ways which create value. If there were other ways to make big money besides unfair information asymmetry, then skilled people would find it easier to do that instead...

Unfortunately, skill/talent does not allow you to gain an information advantage; only social connections do. So the current system is really bad in terms of wasting specialized skills/talents... Current system takes talented people and re-orients them towards adversarial games of deception where they're just a number in a sea of increasingly desperate untalented people who are better at playing the deception game... And the talented folks eventually lose their talents through lack of practice; they're so focused on learning ways to trick people and algorithms, that their actual talents atrophy; by the time they become well known, if they ever do (most don't), they've become mediocre; corrupted by their participation in the process of trying to become relevant; their main talent always takes the back seat.

This is true. Math is more abstract. Some aspects are more tangible; this is what they teach at school and undergrad university... But once you get to master level and PhD, it tends to become increasingly abstract... The utility is only visible to PhD engineers in very specific areas and in those cases, even the mathematicians themselves can't fully grasp the utility of their work and they probably don't really need to.

They don't seem to have any desire to reconcile their 'craft' with real world applications and this is probably why they're particularly good at it.

Yep, about 10 years ago, the security of most major systems was relatively robust; I could not have hacked them. The people who worked there knew much more than I did and they didn't leave many gaps.

Now I could hack essentially any system I want. At least DoS or crash them for sure, with minimal computing on my end. They're much more complex than they used to be. Security-through-obscurity used to be a no-no and sometime in the last 10 years it became the main security paradigm.

You're right, I was typing each term as it came to my mind. Helm chart is the correct term but DevOps people would know what I meant.

ETag is the HTTP response header which provides the shasum (hash) of a resource in order to allow the client to check if the resource has changed without having to load the resource. It's for client side (browser) caching.

SAML is Security Assertion Markup Language; an XML based SSO (Single Sign On mechanism).

CSP (Content Security Policy) which allows the application to specify additional security constraints for the browser to enforce.

BGP; border gateway protocol.

Haproxy is a load balancer.

Helm file; another word for helm chart.

SOCKS5 is a tunnelling protocol which allows you to tunnel through a machine over SSH; you can use it to browse the net or access remote services via another machine (hiding your IP)... It's a bit like a basic VPN. Very easy to setup, you can configure your browser (e.g. Firefox) to access the net via a SOCKS5 proxy so if the remote instance is in a different country, websites will treat you as though you are in that country... Just like a VPN except you have to control the remote machine yourself and log into it via SSH with the -D flag.

Lamport OTS is Lamport One Time Signature algorithm.

Winternitz serves a similar purpose but different tradeoffs and smaller signature size.

SPHINCS is a stateless hash-based signature scheme which works by building a tree of OTS keys (Lamport, Winternitz or other) which can be generated deterministically, on-demand.

Lattice-based cryptography is real.

pg-vector is a plugin for Postgres for vector embeddings and it provides some operators for finding records based on the nearest vector.

sameSite and httpOnly are cookie security settings.

They could drag it out by telling stories of situations they faced but information density would be low and repetitive. As someone who cleaned boats professionally for a year in my early days, I'm pretty sure I could convey all I know about cleaning in under an hour. That's very different from one year of back-to-back lectures.

Cleaning boats, the learning tapered off after a few weeks. With software, I'm 15 years in and it barely tapered at all and it's much more intense. I've been pulling nights and weekends too.

With software knowledge, it would be high information density with little to no repetition. Every piece would provide useful concrete knowledge which would serve to increase technical capabilities and/or security.

>> Cache, stack, heap, process, thread, socket, file, tcp, http, tls, websocks, socks, soc2???, deadlock, stack, queue, race, atomic, event loop, coroutine, async, database, transaction, index, replication, sharding, consistency, serialization, DNS, load balancer, container, namespace, and so on.

... communication protocol, method signatures, web components, ssh, CSS media queries, HTTP headers, WebSockets, timeouts, ETag, iterables, async iterables, middleware, CI/CD, build, consistent hashing, signatures, JWT, SSO, OAuth, SAML, XML, YAML, JSON, block cipher, ETL pipeline, SQL, SQL transactions (atomic), relational databases, foreign keys, schema normalization, referential integrity, 1-to-1, 1-to-n, n-to-n, document databases, compound indexes, idempotency, offset-based pagination, cursor-based pagination, P2P, Kademlia, structured vs unstructured network topology, message routing, frontend router, message storm, reconnect storm, locality, encapsulation, cohesion, coupling, design patterns, modularity, Big O notation, raytracing, shaders, VPN, VPC, data schema, schema validation, CORS, preflight-requests, CSRF, ASCII, UFT8, CSP, CPU context-switching, BIOS, bootloader, interrupt controller, ports, BIND protocol, BGP protocol, assembly language, big endian, little endian, register, signals, embarrassingly parallel, serial processing, event loop, binary trees, tree rebalancing, graph traversal algorithms, sorting algorithms, string character escaping and encoding, blob, base64, UUID, timestamp, CLI, Bash, unit tests, integration tests, e2e tests, TDD, stateful, stateless, proxy, nginx, haproxy, config, helm file, k8s, staging, git, push, commit, merge, rebase, cherry-pick, pub/sub, diff, honeypot, buffer overflow, pointer, file descriptor, authentication, certificates, TLS certificates, DNS Zone files, A record, CNAME, TXT record, SMTP, POP3, SOCKS5, Sha256, HMAC, Merkle trees, Merkle Signature Trees, Lamport OTS, Winternitz OTS, SPHINCS, lattice-based cryptography, pg-vector, vector embeddings, API, rate limiting, cookies, sameSite, httpOnly, localStorage, XSS attack, SQL injection, fetch API, module preloading, bundling...

Barely scratching the surface. I think I could probably keep typing all the technical terms I know for at least 24 hours straight. For most of the topics above, I could probably give a 1 or 2 hour lecture on each one from memory. Some I could give a day-long lecture each.

To explain all the terms I know to a basic degree, I would probably need to give a whole year of lectures back-to-back from 9am to 5pm. And I'm just a rank-and-file senior engineer with 15 years of experience.

It's also why the vast majority of software systems are insecure. The average senior software engineer doesn't know everything that they need to know to build secure software. Last time I poked around Coinbase APIs on HackerOne, I found a DoS vulnerability in less than 30 minutes. That's Coinbase, not some startup built by a bunch of recent graduates.

AI cannot avoid vulnerabilities either since it is trained on average engineer code. There's not enough high quality code available on the entire internet to train AI to implement secure code IMO. As impressive as Mythos may be, it's not enough. I don't even think formal verification would provide protection since sometimes issues with the spec itself can provide an opening for a vulnerability.

I don't know about other people but certainly not for me. I was actually earning crypto passive income for 3 years not working and I was very happy. Now that income is gone, through no fault of my own (literally this was the best case scenario how it could have played out), and I now have to work for money again and I'm very unhappy.

I'm even more unhappy now because my life expectations are higher and I can't get to that level again until I'm 65...

Hearing people tell me this nonsense about money and happiness makes me even more miserable because my conclusion is "They don't appreciate money, yet they are given money... I very much appreciate money, yet I'm not given money..."

From my perspective, it seems almost like the system is only giving money to people who are incapable of appreciating it... Meanwhile I know from past evidence how I felt when I had money + time and that feeling was happiness. I knew it then, I know it now.

Money definitely buys happiness for me. But I resent the amount of competition and how much bs I have to put up with to get it. It's definitely eroding the happiness. Especially when you engage in all the bs for years and still don't get much more money.

For most people, freedom is the only missing piece holding them back from being happy so that distinction melts away for them.

Unfortunately, AI doesn't just automate what we don't want to do. It also automates many of the things which we very much want to do and sometimes pride ourselves on...

And TBH it seems to be automating more of the things we want to do than those we don't... Hence the reason why AI is seen more as additional competition than as additional help.

LLMs are not SaaS. Some LLM are delivered as SaaS. Anyone who has a machine big enough to run a frontier model can launch their own LLM SaaS tomorrow and it would be functionality indistinguishable from any other which is running a similar model...

Also, big companies can choose to run their own models on their own hardware and get better security and privacy as the data doesn't need to leave their own premises.

I want to see it that way too but it seems they've been winning for a very long time with this approach. So is it short-sighted?

The way I see it there is a deep systemic issue beneath this which makes this profitable. That's the flaw which has to be solved.

Private entities should not be able to profit from taxpayer money on such vast scales. Military contacts are massive, foreign aid is massive. It takes from a large number of public hands and gives to a small number of private hands. Diffuse harms, concentrated gains.

I hope you're right though and those issues will be fixed.

The goal is to sell weapons during the war; then afterwards, sell loans to rebuild what the weapons have destroyed... And sometimes it's also to control access to natural resources or trading routes. They don't care who it is who facilitates them on the other side of the ocean; be it a dictator or an 'elected' official.

The lobbyists running the show in the background make money from both the destruction as well as the reconstruction. What is stupid about that?

The politicians later get appointed to lucrative executive roles at the companies which they helped enrich on taxpayer dollars (from both countries). Not stupid either. It's smart retirement planning.

The government is a lot of things but absolutely not stupid; fooling people into thinking it is incompetence is likely part of the ruse... It's a lot easier for people to sit back quietly and watch incompetence unfold, then it is to sit back and watch evil unfold. "It's incompetence" is also a nice easy answer which prevents people from having to look deeper as I have done here.

Julian Assange said about the war in Afghanistan: "The goal is to use Afghanistan to wash money out of the tax bases of the US and Europe through Afghanistan and back into the hands of a transnational security elite. The goal is an endless war, not a successful war."

If you look at any war, you will realize that the same principles are at play. Our current global monetary order has a built-in mechanism for monetizing destruction, unfortunately. It's architected that way, intentionally or not.

I've dedicated much of my life so far to try to convey this simple message.

Not to mention that they're now competing with fake AI slop videos. People don't have so much interest to record their sporting achievements anymore since the video no longer serves as social proof.

> That loss is real and it's worth naming

Yep classic Claude-ism.

The fact that this article was likely AI generated is the real load-bearing factor in this discussion. Or, as previous versions of Claude would say; it cuts through the heart of the issue.

The creative part was only the subject matter. The idea of making a video around "the 3 gifts" is itself kind of creative but fine-grained creativity has been lost.

It reminds me of the meme which shows a before and after of different company logos from like 15 years ago vs now... And now all the company logos look essentially the same; very similar fonts and colors.

People are afraid of creative designs because they are terrified of being judged. People don't want to step out of line.

It's like in a company environment; there are often many unspoken rules and taboos; if someone breaks them, they fall from grace rapidly.

These rules and taboos exist to preserve the structural integrity of the organization. A fragile organization needs lots of rules and taboos to be in place in order to remain cohesive. The people who lead these organizations create the rules and taboos which will allow them to stay in charge. People who don't adhere to the unspoken rules are quickly removed, often with the implicit consent of all group members who are themselves captive to the same rules and taboos and who eventually learn to pride themselves on compliance.

Fully agree. Anyone can get the signals wrong. I guess you should watch how other employees (who have been around for longer) interact with the boss. If everyone is always praising the boss and never criticizing, there's a reason it's like this.