All of this performative bullshit coming out of Anthropic is slowly but surely making them my least favorite AI company.
We get it guys the very scary future is here any minute now and you’re the only ones taking it super seriously and responsibly and benevolently. That’s great. Now please just build the damn thing
These are economic studies on AI's impact on productivity, jobs, wages, global inequality. It's important to UNDERSTAND who benefits from technology and who gets left behind. Even putting the positive impacts of a study like this aside - this kinda due diligence is critical for them to understand developing markets and how to reach them.
But the thing is that they really aren't rigorous economic studies. They're a sort of UX research-like sociological study with some statistics, but don't actually approach the topic with any sort of econometric modeling or give more than loose correlations to past economic data. So it does appear performative: it's "pop science" using a quantitative veneer to push a marketing message to business leaders in a way that looks well-optimised mathematically.
Note the papers cited are nearly all ones about AI use, and align more closely with management case studies vs. economics.
I learned that every step that can be solved reasonably without an LLM, should be solved without an LLM. Reliability, cost, performance, etc.
I try to transfer as much work as I can out of LLMs and into deterministic steps. This includes most of the “orchestration” layer which is usually deterministic by nature.
Sprinkle a little bit of AI in the right places and you’ll get something that appears genuinely intelligent. Rely too much on AI and it’s dumb as fuck.
Make their tasks very small and simple (ideally, one step), give them only the context and tools that they need and nothing else, and provide them with feedback when they inevitably mess up (ideally, deterministically), and hope for the best.
One thing that’s always missing from these compound interest calculators is multiple assets with different rates, and different rates over time (e.g between X date and Y date use Z rate, etc). I didn’t quite figure out the right UI for the second one.
You said “best results”, but I imagine that the theoretical “best” may not necessarily be the most aesthetically pleasing in practice.
For example, limiting output to a small set of characters gives it a more uniform look which may be nicer. Then also there’s the “retro” effect of using certain characters over others.
> limiting output to a small set of characters gives it a more uniform look which may be nicer
And in the extreme that could totally change things. Maybe you want to reject ASCII and instead use the Unicode block that has every 2x3 and 2x4 braille pattern.
News, influencers, Wikipedia, almost all information we consume nowadays is intentional. And not even getting into billions poured into American colleges by the same people.
When I was working in the heart of conservative online media in West Palm Beach—nestled between Rush Limbaugh’s studio, Mar-a-Lago, and Newsmax—targeting Evangelical Christians in the Bible Belt, my salary (and the direction things eventually went) was being paid for by the Saudis. At the time, the propaganda was mostly “pro-oil” and “climate change is a hoax.” Around that same period, those same Saudis bought a 10% stake in Fox News and helped shape the narrative for millions of Christians who tune in and treat it like their main source of news.
So yeah, if you were ever curious where the profits go every time you fill up your car with gas… there.
I thought I was just building media websites. I didn’t even see the content until after six months. I put in my one month notice, finished what I was working on, and left. The amount of money they offered me to stay was ridiculous. I don’t blame people at Fox News for bending the knee and taking that Saudi money -- I just couldn’t make myself do it.
“Thou shalt not bear false witness against thy neighbor.” A lot of people are going to spend eternity in hell for propaganda and lies.
Saudis are invested in a huge range of things in America. It has historical roots in the petrodollar. The basic deal of it that that they would only sell their oil in USD, which gave the USD a de facto backing after we defaulted in Bretton Woods (which was a de jure gold backing). That gave the USD a huge chunk of stability and in exchange we agreed to sell them weapons and broadly support them, while in exchange they were also asked to purchase US treasuries and assets with surplus revenues.
Over time this led to Saudis being involved in just about everything. For instance the biggest owner of 'old Twitter' under Dorsey was Alwaleed bin Talal Al Saud. Needless to say the zeitgeist on old Twitter and Saudi Arabia have basically nothing in common, so you're probably seeing ideological motivation where the real motivation is generally just monetary. Not every country is conspiring to subvert other countries to their ideology.
Basically Saudia Arabia is filthy rich because of oil, but they fully understand that even if we continue burning oil until we run out, we will run out, within the lifetime of some people living today. So they have to migrate their economy away from oil and, on the timeline for such a revolutionary shift, they have very little time left. This is likely what MBS sees as what will define his legacy.
Because I believe in understanding, forgiveness, and redemption.
I have a responsibility not to lie and kill, as commanded in the Bible. I also have a responsibility to tell people not to lie and kill, as commanded in the Bible.
At the same time, our understanding of the science of the mind, as described by Carol Dweck in "Mindset", is that people are not fixed and can change. That is why understanding, forgiveness, and redemption matter. They are essential for helping other people through the process of repentance -- the changing of a mindset.
Can you not understand, forgive, and believe in redemption, but also judge?
"I understand why you took oil money from the royal family famous for murdering journalists; money is nice to have. However, I judge you for it and will not associate with you until you redeem yourself through seeking forgiveness and changing your behavior."
I was introduced to Maslow's hierarchy of needs 25 years ago around the same time I read about Pavlov's dogs.
The need to belong is extraordinarily motivating. It became obvious that the cults leveraged the need in the individual to belong to a group by accepting the person without judgement first rather than attacking the person they are trying bring into their group pushing them away.
I get told this a lot by liberals, that it's wrong that I shouted in a cop's face that he's a fascist pig and a traitor to the people, now he'll never support my cause, but I'm not really sure I agree. The cop, and the nazis he's protecting from me, will never join "my group" in ten million years, no matter how nice I am to them. Do you believe otherwise?
Yeah I get what you mean, but it’s not really about converting the cop into “your group.”
It’s about what your actions do to everyone watching, and what it does to you.
Plato makes the point that you don’t make a dog better by beating it. You just make it worse. Same with people. You’re not persuading, you’re escalating.
If the goal is change, you don’t have to be nice, but you do have to be effective!
Well, let's explore the topic then, because for example aforementioned cults will use protest or other uncomfortable situations to solidify indoctrination. See: Mormons and Jehovah's witnesses sending people door to door to proselytize knowing full well the majority of people will be annoyed by this, which will Other the proselytizers and make them feel like the church members are the only people who they're safe around. Or the "God Hates F*s" Church doing their protests. Taken to the extreme: the Cultural Revolution's struggle sessions, designed explicitly to make as many people as possible feel that they were culpable alongside the Party. So, maybe not great for the opponents or the observers, but very good at solidifying the base itself.
Personally I'm not interested in running a cult, but I'm very interested in anything that empowers people.
In the case of an anti ICE protest where we shout mean things at the gestapo, a couple side effects include the empowerment of participants and locals. See for example how the dynamic shifts for the woman sheltering a door dash driver from ICE once more neighbors start showing up: https://www.reddit.com/r/minnesota/comments/1q8vvwa/st_paul_...
In the case of the various anti Nazi protests I've been to (proud boy rallies) it's also been good for generating images of just how many people are in opposition to racists.
It's not necessarily always about persuading, sometimes it's more about, well I suppose "circling the wagons?" Solidifying community support, demonstrating capabilities, empowering people and communities, and disempowering, defanging, or scaring racists and fascists. Finally, it's great for recruitment: fed up liberals turn up to their first protest, get one hell of an adrenaline rush screaming at cops and running away from tear gas, and then may later ask the person pouring milk onto their face how they can help outside of protesting. In that sense the cop's escalations, while barbaric and inexcusable, are the unconvincing escalation you mentioned that in fact helps us.
But for you then, I'm not sure your opinions on ICE as gestapo but perhaps humor my position on their danger, how would you instruct anti-fascists to operate in the USA right now in the face of ICE raids? The original idea is, what, applying Christian values? Jesus threw out the merchants and moneychangers, did he not? For certain people, he decided he wasn't in the business of forgiveness.
For what it's worth, I generally agree with what you're saying, my goal in conversation is always to just pull people left. I just have a practical and situationally pragmatic limit.
Looks like Prince Alwaleed bin Talal's ownership of Fox was between 5.5% to 7% during the two-decade period of 1997 ~ 2017. He divested during an anti-corruption purge.
What level of moral compromise is acceptable in this world to take whatever money is offered? Presumably the job of hitman is unacceptable? Where's the line drawn?
Personally I'd say that lying to perpetuate a system that is leading to various populous parts of the world becoming uninhabitable is on the wrong side of that line.
There are agendas there, just like in every human endeavor, but it definitely hasn't been "hijacked", it's still by far the best single repository of human knowledge out there. If I had to choose one website to take with me to a desert island, it's an obvious choice.
We should keep talking about the issues and improving things, but don't throw out the baby with bathwater.
> We should keep talking about the issues and improving things, but don't throw out the baby with bathwater.
Yeah, I wonder what solution people propose that claim that Wikipedia is 'hijacked' or 'compromised' and pushing agendas?
While Wikipedia is not perfect, it is the best encyclopedia we currently have, mostly due to collective efforts and maintainers that care about the state of Wikipedia. I would even say that it is a good thing that there is this transparency, that states and capital are trying to influence Wikipedia because then you know that you may take some articles with a grain of salt or can actively push against it.
Every alternative to Wikipedia that I have seen so far is one that claims to be more truthful than the original, but in the end these are platforms that push agendas without the transparency and attempt to further obscure power relations under the pretext of truth.
Every alternative to Wikipedia will have to solve the problems that Wikipedia already has to be a better alternative. However, I do think these are fundamental unsolvable problems
and everyone who claims to have solved this is part of a power struggle over who defines what is considered true.
More transparency around the admins and the hierarchy above it would be a good start, as would some kind of countervailing pressure to the ballooning of meta rules (bylaws). For instance:
- Oppose the "Super Mario" effect: if admins do something ordinary users would get banned for, they get banned too, they don't just lose their admin title.
- Implement restrictions on Arb Com to make it worthy of its "supreme court" moniker. Provide prior notice, allow representation, access to evidence ahead of the case, and require the Arb Com to disclose the logic of any automated scripts they use for mass judging (e.g. counting proportion of edits being reverts, or that counts every change to a reference as "reference vandalism"). Grant defendants the ability to force the Committee's judgment to be disclosed to the public, with PII redacted if necessary.
- Require that precedent be recorded for unclear meta rules: what counts as a violation of e.g. canvassing? When do reversions become evidence or proof of "ownership"?
- Create an independent appeals body for Arb Com decisions. Like the Arb Com itself, the logic or source code for any scripts they use to aid their decisions, should be public. Ideally, choose the independent appeals body by different means than the Arb Com itself is chosen, e.g. by random selection of users with a certain activity level, independent of the ordinary admin track.
- Grant all users the right to be forgotten (courtesy vanishing), not just users in good standing, so that users bullied off the platform can remove their proverbial stockade.
- Create a mechanism that forces rules to be refactored or reduced in scope. Just spitballing, one possible way might be to limit the growth of any given WP: page per unit time, require negative growth for some of them, or in some way reward editors who reduce their extent.
There may be fundamental unsolvable problems, but that doesn't mean the current system can't be improved.
It is a great example of the shaping of opinions the OP claims Wikipedia suffers from. It is a textbook example of the way the detractors of Wikipedia comport themselves.
Accuse the site of of exactly what you’re doing at this exact moment.
it absolutely has been. like every online community, Wikipedia is extremely vulnerable to the terminally online and/or the mentally ill, to whom everything is political. like clockwork, every remotely political article cites opinions only from a certain perspective, often quoting glorified nobodies to assert the narrative the '''editors''' want to present. dissenting opinions, no matter how overwhelmingly common among the real people, are mentioned in passing at best and often derisively.
I went to Wikimania in London and the community who turned up were pretty middle of the road types. Mostly retired males doing after it they'd finished with their prior job in a variety of fields.
For example, this article goes over disinformation by polish nationalists on Holocaust related articles. There's a chart with 10 editors accounting for 50% of the edits and another chart that shows disproportionately citing authors that in reality are not academically are under-cited
Extraordinary claims require extraordinary evidence. That Wikipedia has been co-opted by mentally ill people is an extraordinary claim. You should provide more than feelings.
Wikipedia's model does favour autistic* people and those with a lot of time on their hands. You can see this in the sheer volume of some contributions, their focuses and the invention of obscure rules and Wikipedia specific jargon e.g. peacock terms etc.
* ASD is not a mental illness but it can produce quirky and obsessive behaviour.
The Scottish Gaelic Wikipedia was very much hijacked. As was the Lowland Scots one, except in the second case the individual made the news. The Gaelic one has a German guy running it who has several vanity articles and has chased most other users off apart from some government employees.
I think for anything controversial we need a completely different model.
Officially wikipedia is NPOV but an especially contentious and murky political mudfight decides what counts as a "citeable" source and what doesnt and what counts as notable and what doesnt.
It also has an incredibly strong western bias.
Every government, corporation and billionaire pays somebody to participate in that fight as well, using every dirty trick they can.
Until we have a model that can sidestep these politics (which Wikipedia seemingly has no real desire to do) and aggregate sources objectively I think it will continue to suck.
I agree with the issues, but it definitely doesn't suck if compared to every single other massive endeavor out there. As I see it, it's like that quote about democracy - it's the worst way to attempt to catalogue human knowledge, except for all those other forms that have been tried.
> It also has an incredibly strong western bias.
What's the issue with that? Why shouldn't English Wikipedia have a strong Western bias? I've explored and participated in several other Wikipedias and other collaborative projects, and each is biased towards the worldviews common to the culture that its main editors come from. I don't think there's a way to have an encyclopedic project without any cultural bias at all (if such a platonic ideal could even be properly defined), and seeing how Western values include a significant focus on pluralism, freedom of expression and scientific inquiry, I think this situation is much better than the alternatives.
>I agree with the issues, but it definitely doesn't suck if compared to every single other massive endeavor out there.
Compared to what? I dont really see much aggregation being done at all on Wikipedia's scale.
>What's the issue with that?
It's supposed to be impartial and objective and it sells itself as such but if you see how the sausage is made it is patently the exact opposite.
>I don't think there's a way to have an encyclopedic project without any cultural bias at all
I think it's perfectly possible to have an encyclopedia which is more liberal about allowing more sources to be used and which provides tools and metadata about those sources and gives tools to the user allowing them to filter accordingly.
Whereas "Blessing" one group of sources and condemning another will inevitably turn it into a propaganda outlet for whomever controls it. You might think that this is the only way but that represents more of a failure of imagination than a lack of options.
The way some billionaires are described on Wikipedia you'd think they were saints. Even though most of their philanthropy is a tax write-off and goal-orientated (producing good publicity for them or pushing society in a direction they want).
But by claiming one thing and doing the exact opposite (on a statistical quantitative basis), Wikipedia and all other western outlets have become just a front for propaganda which is also the reason why I don't believe in "Persecution of Uyghurs in China"
German Scholars Reveal Shocking TRUTH About China’s Xinjiang Province
That video's main gripe seems to be that western media gloss over things being kicked off by islamic terrorists but from the Wikipedia:
>...Uyghur terrorists killed dozens of Han Chinese in coordinated attacks from 2009 to 2016. These included the September 2009 Xinjiang unrest, the 2011 Hotan attack, the 2014 Kunming attack, the April 2014 Ürümqi attack, and the May 2014 Ürümqi attack. The attacks were conducted by Uyghur separatists, with some orchestrated by the Turkistan Islamic Party (a UN-designated terrorist organization)...
"The US government hates China and also hates Muslims but suddenly cares about Muslims in China" - forgot who quoted this. Compare US stance on Gaza. If media can ban discussions on gaza, can't it invent phantom genocide in Xinjiang Province? Something to think about.
Also let's not forget "Persecution of Uyghurs in China" use to be titled "Uyghur Genocide" or some variation thereof for years. Never mind plurality of UN countries has recognized PRC actions in XJ as counter extremist/terrorists, or even among western bloc most countries did not label it as "genocide", bu somehow wiki and captured editors went with the genocide framing, aka, fucking we lie, we cheat, we steal Mike Pompeo serving as secretary of state for US geopolitical interest designation during sino-us coldwar. Because Wiki NPOV isn't based on reality but "reliable sources" which just happen to be western aligned that parrot each other to manufacture consensus / propaganda. Even "persecution of Uyghurs" still biased considering plurality of world still considers PRC actions in XJ as de-deradicalization / counterterrorism, and the numbers have only swung more in PRC favour over time - geopolitical reality is "Chinese War on Terrorism" whose causalities paled in comparison to wiki's "(Global) War on Terrorism" that would otherwise be characterized as "genocide/persecution of Iraqis/Afghans" which killed and displaced millions. Wonder if Obama would have gotten a Nobel Peace prize if that article title existed.
internet altered the way society communicates and why, a lot of discussions now end up by "show me your sources" aka "what is the truth" and it's often centralized into some accepted source like wikipedia
where there simple single point of 'truth' like that before ?
my 2cents is that humans are not meant to live in one global absolute truth and we all lived in relative fuzzy reality before, it was slow and imperfect but not as easy to tamper with
Showing sources is not a bad thing. The harm is not questioning sources. A lot of people rely on poor sources. Whatever what the first result in Google historically, and now LLM summaries.
maybe my viewpoint is weird but i think this distorted human interactions on multiple domains.
of course we all wanted to communicate faithful information, but now any discussion turns into a religious difference, and the escape is of course "who has the truest source". people don't necessarily understand the content, they just defer the validity to an official third party, so basically we're back to zero.. but we're all debating everything now.
and it makes me think that locally, we chatting, was never meant to exchange rigorous information, but mostly to share opinions lightly, more emotional than rigorous and scientific
We still live in that fuzzy reality. Not much has changed.
It doesnt really matter if the whole world has access to the same information if the whole world trusts completely different sources.
For better or worse we trust those sources exclusively because of tribal affinity.
I doubt many people in the US could be persuaded to trust Global Times over the New York Times even if you could prove it had a better prediction track record. Wrong tribe.
And now it's given by the mainstream media, which is mostly owned by a few very rich people and pushes the same type of propaganda as before (but now globally).
There was a point where I would agree with that, but we seem to be moving past that. The "truth" seems to be coming more from social media influencers than mainstream media now.
> Colleges are political, and donations are made to assure they keep on being political.
Ok, to clarify:
> > And not even getting into billions poured into American colleges by the same people.
Which American colleges, by what people, and what does "being political" mean? Maybe I'm very ignorant of the USA, are these just known things to Americans?
You need only to look at how many actual well credentialed doctors get their Wiki pages smeared with words like "misinformation spreader" for dissenting against covid narratives
It seems today that he was just wrong and used to make "dubious" clinical trials.
> As of 2025, 46 of Raoult's research publications have been retracted, and at least another 218 of his publications have received an expression of concern from their publishers, due to questions related to ethics approval for his studies.
Raoult's case is so strange.. he's not the usual fringe doctor, up until covid, he had a center seat in national health institution and everybody around him was listening. I still don't get why nobody was wary of him there..
> Anyone with two braincells could see it at the time.
It seems a captain obvious now but it wasn't so at the time. (Or maybe my.braincells.count() < 2)
Many people listened because he wasn't some youtuber doing his research, he was the head of the "Infectious and Tropical Emergent Diseases Research Unit" ad the Faculty of Medicine of Marseille.
I've watched one of his interviews where he stated that people survived in his unit with hydroxychloroquine and that he had numbers to prove it.
When you look at his credentials, and my.braincells.count(), it was hard to identify it as misinformation.
I definitely exaggerated with my "two braincells". Even the french president said about the guy "we need more people like him" (although I wouldn't say he's that smart himself...).
But even without being knowledgeable about statistics, there were a lot of very serious people giving very good arguments against his results. You just had to see them. And seeing all the Facebook doctors lunatics instantly side with Raoult and defend him tooth and nail should definitely raise some red flags...
They probably mean people like Robert Malone [1], who - despite being well accomplished in a related field - spread verifiably wrong information about vaccines on social media during the pandemic. There are many people like him who showed past accomplishments in a related field, but were totally out of their depth when interviewed about covid on the Joe Rogan podcast or similar.
Yet in officialdom, that kind of thing was perfectly acceptable. In Scotland we had a dentist running Covid lockdown, which is ironic since public dental services were decimated by it and never recovered.
You can simply do a Wikipedia search for "misinformation doctor" and get plenty of results, even with its search system, let alone if you use a search engine to power the search.
I would think that posting any particular person would descend in to a pointless argument over whether those claims are merited. Do you have some better reason to want a particular name?
If there is misinformation on Wikipedia it can be corrected. Unless you are claiming that all hits for "misinformation doctor" are incorrect, a few examples to verify and correct would be helpful.
I can point you to several pages that are protected by groups of interested admins that will make changing even blatantly obvious misinformation impossible, let alone contentious stuff.
Have you ever tried changing something on Wikipedia regarding politics (which now includes several health issues) or religion?
Edit: also, I did write "I would think that posting any particular person would descend in to a pointless argument over whether those claims are merited." and yet you're suggesting I get into that argument. I quite clearly don't want to because it's pointless, and we had years of it anyway.
> I can point you to several pages that are protected by groups of interested admins that will make changing even blatantly obvious misinformation impossible, let alone contentious stuff.
Some 'misinformation' is hard to correct because the corrections are reversed by those who are intent on spreading the 'misinformation'. This is especially prevalent around contentious and/or politically sensitive subjects like the mentioned SARS2-related cases. This is what makes it hard to trust articles on such subjects on Wikipedia.
If this is quite widespread, it should be fairly straightforward to point to an example of a page that's being defaced with misinformation, which would include an edit history and perhaps a Talk page documenting whatever sides to the debate there is that's preventing consensus.
I don't disagree that weird bullshit occasionally happens on Wikipedia, but I have noticed that as soon as light is cast on it, it usually evaporates and a return to factual normality is established.
My go-to example is the "Constitution" of Medina[1]
> It is widely considered to be one of the first written constitutions of mankind.
Now go to the page on constitutions in history[2] and see how far down the list that one is.
Now go back to the Constitution of Medina (itself an example of misinformation, since it should be charter or even more precisely, treaty, but those protecting the page have meddled with the title too) and look at the reference it uses[3] and what it says to get a feel for the kind of "reference" that is being used there, and then try and update said Wikipedia page by removing the parts about its being the first.
The talk pages of both show that invested groups have been trying to force their views, and they've done it quite successfully.
Let us all know how you get on with that, and then I'll point you to the next example, and the next example…
Some other notable things to check are co-founder Larry Sanger's 9 theses[4], and the news that broke yesterday about a PR firm doing "Wikilaundering"[5].
I don't really understand how you've come to that conclusion. If you look at the protection log[1], Constitution of Medina was protected in 2016 for a bit under a month, and never outside of that. The "earliest constitution" was also discussed in 2016[2][3], and there was consensus not to include the claim. Then, in November 2025, it was re-added by a new editor who made no other edits[4].
Looking at the talk page of Constitution, it was discussed exactly once, in 2005[5].
>> those protecting the page have meddled with the title too
> Why is this "consensus not to include the claim" relevant when the claim was already included?
Because anyone can dispute anything. But saying it's some kind of agenda by a group of admins is incorrect.
You’re taking those questions too literally. The need for dispute resolution implies a dispute, well done… if you’re in to one-step thinking. Explain how there was a dispute over the facts there and how it wasn’t intentional misinformation pushed by a group of interested parties that have continued to press their case from before that date until now.
Or, you can put it down to an honest mistake or difference of opinion. That really is the oldest written constitution in the world, or it’s got a valid claim to be, and those people don’t want to add any respectability to their pet project.
Tough choice. The phrases “die on that hill” and “never interrupt your opponent when they’re making a mistake” come to mind. Do continue.
I cannot fathom where you get "intentional misinformation pushed by a group of interested parties". You're welcome to read the original dispute at [1]. Such things are not uncommon when collaboratively editing. There doesn't need to be a cabal of editors behind it.
This must be one of the more bizarre conspiracy theories I've heard.
Again, please explain how such an obvious piece of misinformation wasn't misinformation but an honest mistake, yet occurring over several years and with several people, some of whom were sock puppets and still it persists in some form.
"Arguing in bad faith" - what would that actually mean? Would it be the same as using a sock puppet to push an agenda? That wasn't me, that's what I'm pointing out and you're dismissing for no good reason.
Regardless:
- The page is still titled "Constitution…" when the opening paragraph contains "The name "Constitution of Medina" is misleading as the text did not establish a state." Make that make sense.
- "and the first "Constitution"" is still in the page
It persists.
Now, what I might consider bad faith is:
- being unwilling to answer simple, straightforward questions, which is apt, considering Socrates was an Athenian
- having such an interest in the page that you claim you made edits
- not checking properly and thus thinking this only happened twice, and wasn't part of attritional arguments, rollbacks, edits and counter-edits
Wikipedia must be alright if one does not wish to see a problem.
If your review was based on features shipped, and your bosses let you send PRs with no tests, would you? And before you say "no" - would you still do that if your company used stack ranking, and you were worried about being at the bottom of the stack?
Developers may understand that "XYZ is better", but if management provides enough incentives for "not XYZ", they're going to get "not XYZ".
That actually wasn't why I didn't write tests a lot of the time.
What stopped me was that after a year of writing tests, I was moved to a higher priority project, and the person who followed me didn't write tests.
So when I came back, many of the tests were broken. I had to fix all those in order to get new ones to not be a bother.
Repeat again, but this time I came back and the unit testing suite had fundamentally altered its nature. None of the tests worked and they all needed to be rewritten for a new paradigm.
I gave up on tests for that system at that point. It simply wasn't worthwhile. Management didn't care at all, despite how many times I told them how much more reliable it made that system, and it was the only system that survived the first giant penetration test with no problems.
That doesn't mean I quite testing. I still wrote tests whenever I thought it would help me with what I was currently working on. And that was quite often. But I absolutely didn't worry about old tests, and I didn't worry about making sure others could use my tests. They were never going to try.
The final straw, less than a year before I was laid off, was when they decided my "storybook" tests weren't worth keeping in the repo and deleted them. That made me realized exactly how much they valued unit tests.
That isn't to say they had no tests. There was a suite of tests written by the boss that we were required to run. They were all run against live or dev servers with a browser-control framework, and they were shaky for years. But they were required, so they were actually kept working. Nobody wrote new tests for it until something failed and caused a problem, though.
tl;dr - There are a lot of reasons that people choose not to write tests, and not just for job security.
Well this wasn’t really aimed at individual devs, but the team/company standards.
I’ve worked at several teams and it was always the norm that all PRs come with tests. There was never a dedicated QA person (sometimes there would be an eng responsible for the test infra, but you would write your own tests).
I would never accept a PR without tests unless it was totally trivial (e.g. someone mentioned fixing a typo).
Depends on how easily the failure is connected back to you personally. If you introduce a flaw this year and it breaks the system in two years, it won't fall back on you but the poor sap that triggered your bug.
It depends on the application but there are lots of situations where a proper test suite is 10x or more the development work of the feature. I've seen this most commonly with "heavy" integrations.
A concrete example would be adding say saml+scim to a product; you can add a library and do a happy path test and call it a day. Maybe add a test against a captive idp in a container.
But testing all the supported flows against each supported vendor becomes a major project in and of itself if you want to do it properly. The number of possible edge cases is extreme and automating deployment, updates and configuration of the peer products under test is a huge drag, especially if they are hostile to automation.
Once, for a very very critical part of our product, apart from the usual tests, I ended up writing another implementation of the thing, completely separately from the original dev, before looking at his code. We then ran them side by side and ensured that all of their outputs matched perfectly.
The "test implementation" ended up being more performant, and eventually the two implementations switched roles.
To put things in context, it both depends on organization standards, and what the change actually is.
Where I work, there are areas that, if you change, you must update the tests. There are also development helper scripts and internal web sites where "it compiles" is good enough.
Likewise, I've done quite a bit of style cleanup PRs where the existing tests are appropriate.
Yes, depends on what you're building. Is it just a prototype? no tests needed. Are you trying to move fast and break things? no tests needed. Are tests just not feasible for this piece of code (e.g., all UI, not unit testable), then no tests needed.
This seems to be where it’s at right now, we can’t seem to make the models significantly more intelligent, so we “inject” our own intelligence into the system, in the form of good old fashioned code.
My philosophy is make the LLMs do as little work as possible. Only small, simple steps. Anything that can be reasonably done in code (orchestration, tool calls, etc) should be done in code. Basically any time you find yourself instructing an LLM to follow a certain recipe, just break it down to multiple agents and do what you can with code.
i have a slightly different but related take. the models actually are getting smarter, and now the challenge becomes successfully communicating intent with them instead of simply getting them to do anything remotely useful.
Gambit hopefully solves some of that, giving you a set of primitives and principles that make it simpler to communicate intent.
reply