I haven’t personally tried it yet. I do certainly battle Claude quite a lot with “no I don’t want quick-n-easy wrong solution just because it’s two lines of code, I want best solution in the long run”.
If the system prompt indeed prefers laziness in 5:1 ratio, that explains a lot.
I will submit /bug in a few next conversations, when it occurs next.
Very interesting. I run Claude Code in VS Code, and unfortunately there doesn't seem to be an equivalent to "cli.js", it's all bundled into the "claude.exe" I've found under the VS code extensions folder (confirmed via hex editor that the prompts are in there).
Edit: tried patching with revised strings of equivalent length informed by this gist, now we'll see how it goes!
I didn't know we could change the base system prompt of Claude Code. Just tried, and indeed it works. This changes everything! Thank you for posting this!
Remember Sonnet 3.5 and 3.7? They were happy to throw abstraction on top of abstraction on top of abstraction. Still a lot of people have “do not over-engineer, do not design for the future” and similar stuff in their CLAUDE.md files.
So I think the system prompt just pushes it way too hard to “simple” direction. At least for some people. I was doing a small change in one of my projects today, and I was quite happy with “keep it stupid and hacky” approach there.
And in the other project I am like “NO! WORK A LOT! DO YOUR BEST! BE HAPPY TO WORK HARD!”
This might be more complex than I imagined. It seems Claude Code dynamically customizes the system prompt. They also update the system prompt with every version so outright replacing it will cause us to miss out on updates. Patching is probably the best solution.
> Given its key importance for the EU's competitiveness, the Commission is calling on the European Parliament and the Council to reach an agreement on the EU Inc. proposal by the end of 2026.
Try asking it for some curves that go in this or that direction. CW, CCW, curves to the left, to the right, etc.
Produces a bunch of crap and it’s really hard to convince it that something needs to be changed. “Yep, I looked at the screenshot, it curves, so it’s okay.”
You can't call filter() without a model. It's always Folder.objects.filter(user_id=user_id), so the context is right there in the code. Plus the generated .pyi stubs give your IDE full type info per model, so "go to usages" works through the type system.
When they’re 30 lines apart, because it’s a function that builds a complex query with joins and filters based on some function arguments: not easy.
I don’t really understand how .pyi will give me proper Find usages here. I haven’t used them a lot, though, and just rely on pycharm Find Usages or just rg.
Also, what about foreign keys? Let’s say it’s a filter on user__email__startswith. Are you able to find it when executing Find Usages on User.email field?
How can they alter humanity? What's the difference for humanity since CERN found Higgs particle? In what ways could the potential dark matter particle detection alter humanity?
It’s a place where extremely skilled people work highly motivated on humanities hardest problems at scale.
CERN pushed distributed computing and storage before anyone else hat problems on that scale.
CERN pushed edge computing for massive data analysis before anyone else even generated data at that rate.
CERN is currently pushing the physical boundaries of device synchronisation ( Check „ White Rabbit“ ), same for data transmission.
CERNS accelerator cooling tech paves the way for industrial super cooling, magnet coils push super conduction…
Companies are always late in the game, they come once there is money to be had:
No one founded a fusion startup until we were close enough to the relevant tripple product.
You are perfectly right, this has been similar to the "space industry" (which includes 'ballistic nukes' knowhow maintainance). The thing with a bigger collider is it seems there are, not that honnest, scientists retro-fitting models in order to reach 'appropriate for this new collider' energy ranges where 'new physics' could be found.
What does that even mean? The FCC is essentially the next plausible energy range we can probe with a collider.
Going larger would cost more, and add risk.
So like, yes? The obvious thing to do is to analyze our models and come up with experiments to do within energy ranges which are plausibly accessible with near future technology.
This is where there is a questionable issue: some network of dishonest scientists may have retro-fitted the models in order to get realitic energy ranges for this new collider.
Sure, but if experimental physics don't matter, wouldn't it be a far better idea to develop all those kinds of technology without actually building the expensive collider itself?
In what way would studying black body radiation alter humanity? Oh just the basis for quantum mechanics and thus transistors, lasers, MRIs, photovoltaics, and more.
The point is, you don't know in advance. I admit it's a bit more far fetched with these experiments that are so far removed from everyday life, but they're still worthwhile.
I understand how linacs and even small compact syncrotrons can have practical medical and industrial applications, and I understand that in the past CERN has developed technology and produced research which is relevant to hardon therapy.
What I don't understand, and maybe you can clarify, is how the very largest gargantuan accelerators can ever have practical relevance. How can effects and products which can only be studied with accelerators that are many miles large ever have application in hospitals unless those hospitals are also many miles large? Not going to lie, I get "NASA invented Tang" vibes whenever this subject comes up; like the medical applications of small accelerators are obvious and parsable to the public, so they are used to sell the public on accelerators the size of small countries.
Because of the engineering effort required to build such systems, that no one has built before, means there is a gigantic amount of R&D discoveries that can be eventually applied in other fields outside particle physics.
Hence why CERN eventually created an industry collaboration office, responsible for finding business partners that would like to make a business out of such discoveries.
The internet existed, hypertext existed, it was just happenstance that it was put together there. It would have happened somewhere, maybe not exactly the same protocol but the same end result.
I disagree that any new possibility for treatments should be lauded. The theoretical side of things is fine, but many new treatments are far more expensive than existing options without offering improved outcomes.
This is orthogonal to your point about CERN being useful.
> Some people also believe praying beats vaccination programs.
> Unfortunately I have got to know people that are only still around me thanks to this technology that you find needless.
There is no way to know whether these people would have been served better by receiving radiation therapy. Your statement is tantamount to believing in prayer.
What do you mean by 'any evidence that works better Than alternatives'?
It can deliver radiations to the brain that will peak at the exact position of the cancer, and reduce irradiation in sane tissues.
The 'better' is 'less irradiation to sane tissues' that in turn reduces the risk for new cancers.
Note: I'm not expert on the matter, but I had technical visits to IBA and know several PhDs that work there
> What do you mean by 'any evidence that works better Than alternatives'?
I mean exactly that, clinical trials demonstrating that proton therapy is superior to radiation therapy. This is not a question about the physics but about how patients respond (and whether the expense of delivering proton therapy outweighs the expected marginal benefits).
But on the subject of discoveries and practical uses, the IBA cyclotrons are also used for other purposes than proton therapy: cleaning exotic fruits from dangerous substances and personalized medicine.
This may be one of the good cases, then. I'm not an expert in cancer but I am a biologist and physician. The head and neck cancer (here) and various pediatric indications get the most attention but it has felt that proton therapy has been seeking an indication for almost 40 years now.
The study was designed to show non-inferiority, which doesn't preclude their ability to show an improvement. It would be helpful to see other studies before determining that proton therapy is better (or even non-inferior) to radiation therapy. It's certainly much more expensive, which shows up in the study as many subjects being denied insurance coverage.
Edit: This is now in the weeds, but the per-protocol participants didn't fare better than the intention-to-treat participants, which one might expect since insurance approval lead to dozens of subjects changing treatment arms.
In Europe at least, many insurances cover it if you have the right criterias.
From my visits, they mostly focus on children that have some very nasty cancers, the IBA hospitals are all designed with children in mind (to avoid stressing them), and from my memory, a unique hospital is often enough to treat a whole country for the kind of cancer they target.
Now, if it is on par with classical radiotherapy BUT it gives less subsequent problems, it might be worth the cost as subsequent problems can be as expensive or even more than the original treatment. It becomes an actuarial issue to know where is the tradeoff.
I think we agree in general, I don't disagree that maybe Proton therapy is not better than radiotherapy, it might but we lack some evidence.
I only argue that if they are equal in quality of treatment and the 'total cost' is the evaluation parameter, it is way more complex than the treatment itself, and it could be justified to use proton therapy, even if more expensive.
Well, when I’m driving in Kyiv, and there is an air raid alert, usually my car navigation starts to derp, and after a few minutes it thinks that it’s suddenly in Lima, Peru.
Not that I mind too much, I know how to get around without navigation.
It does teleport to Peru but it also fast-forwards time to about a year into the future, which caused my car to think its overdue for that oil change. It even synced that back to the headquarters and I got an email asking me to take it to service.. (and arriving there on the wrong side of the Dnieper, I just decided to wait it out)
Wish we could put it into a manual mode where you just reset it's position once and then it updates based on wheel encoders & snapping to roads.
Here is a gist that tries to patch the system prompt to make Claude behave better https://gist.github.com/roman01la/483d1db15043018096ac3babf5...
I haven’t personally tried it yet. I do certainly battle Claude quite a lot with “no I don’t want quick-n-easy wrong solution just because it’s two lines of code, I want best solution in the long run”.
If the system prompt indeed prefers laziness in 5:1 ratio, that explains a lot.
I will submit /bug in a few next conversations, when it occurs next.
reply