Here's an email that keeps things professional and firm while making clear you expect action:

Subject: Follow-Up: Heating Repair Requested Three Weeks Ago – Unit [Your Apartment Number]

Dear [Landlord's Name],

I hope you're doing well. I'm writing to follow up on the heating issue in my apartment, which I first reported on [date] — three weeks ago today. As of this morning, the heating is still not working.

I understand that repairs can take time to schedule, but three weeks without heat in [current month/season] is a significant problem. Beyond the discomfort, temperatures in the apartment have dropped to a point that affects my ability to live there comfortably and, frankly, raises health and safety concerns.

I would appreciate the following:

1. A specific repair date — a firm appointment, not a general promise that someone will "stop by."

2. A temporary solution in the meantime if the repair will take longer (e.g., space heaters or a rent adjustment for the period without heat).

3. Confirmation in writing of the repair plan.

If I don't hear back with a scheduled repair date by [date — e.g., end of this week], I'll need to explore other options to resolve this, including [local housing authority / tenant rights resources] as applicable in [city/state].

I'd much prefer to settle this directly and simply, and I appreciate the good relationship we've had as tenant and landlord. Please let me know the repair date at your earliest convenience.

Thank you, [Your Name] [Apartment Number & Address] [Phone Number]

A few tips to strengthen your position without escalating:

• Send it in writing (email is good) — this creates a paper trail if you need it later.

• Attach or reference your original request ("as I mentioned in my text on Feb 3...").

• Check your local tenant laws — many jurisdictions legally require landlords to maintain heat, and some let you pay for repairs yourself and deduct from rent if the landlord is unresponsive. Mentioning you know your rights (the way the email does, gently) often motivates action.

• Keep the tone factual — dates, specifics, and clear requests. Avoid accusations like "you've been ignoring me."

Would you like me to adjust the tone (firmer or softer), or tailor it to your specific situation?

useDebounce Hook

import { useEffect, useState } from "react";

/**
 * Returns a debounced copy of `value` that only updates
 * after `delay` ms have passed without `value` changing.
 * The pending timer is cancelled on unmount or re-render.
 */
function useDebounce<T>(value: T, delay: number = 500): T {
  const [debouncedValue, setDebouncedValue] = useState<T>(value);

  useEffect(() => {
    // Start the timer whenever `value` changes
    const timer = setTimeout(() => {
      setDebouncedValue(value);
    }, delay);

    // Cleanup: cancels the timer if value changes again,
    // or if the component unmounts
    return () => clearTimeout(timer);
  }, [value, delay]);

  return debouncedValue;
}

export default useDebounce;

Types explained

• <T> generic — the hook infers the type of whatever you pass in, so debouncing a string returns a string, an object stays that object type, etc.

• value: T — the reactive input.

• delay: number = 500 — optional, defaults to 500 ms.

• Return type T — same type as the input.

Usage example

function SearchBox() {
  const [query, setQuery] = useState<string>("");
  const debouncedQuery = useDebounce<string>(query, 300);

  useEffect(() => {
    if (debouncedQuery) {
      // Only fires 300ms after the user stops typing
      console.log("Fetching results for:", debouncedQuery);
    }
  }, [debouncedQuery]);

  return (
    <input
      type="text"
      value={query}
      onChange={(e) => setQuery(e.target.value)}
      placeholder="Search..."
    />
  );
}

Key points

1. Cancel on unmount — the useEffect cleanup function runs clearTimeout(timer), so no state update is attempted after the component unmounts (avoiding the "setState on unmounted component" issue).

2. Re-render safety — cleanup also runs on every value change, which is exactly what makes debouncing work: a rapid typist keeps resetting the timer.

3. Generic type safety — useDebounce<string>(query, 300) or simply useDebounce(query, 300) (inferred) both work.

The Answer: About 33%

Despite the test being "99% accurate," the probability the person actually uses the drug is only ~1 in 3.

The Calculation (Bayes' Theorem)

Given:

• P(uses drug) = 0.5% = 0.005

• P(tests positive | uses) = 99% = 0.99

• P(tests positive | doesn't use) = 1% = 0.01 (false positive rate)

Applying Bayes' theorem:

$$P(\text{uses} \mid \text{positive}) = \frac{P(\text{positive} \mid \text{uses}) \cdot P(\text{uses})}{P(\text{positive} \mid \text{uses}) \cdot P(\text{uses}) + P(\text{positive} \mid \text{doesn't}) \cdot P(\text{doesn't})}$$

$$= \frac{(0.99)(0.005)}{(0.99)(0.005) + (0.01)(0.995)} = \frac{0.00495}{0.00495 + 0.00995} \approx 0.332$$

Intuitive Version: Think in Numbers

Imagine 10,000 people tested:

So about 149 people test positive, but only ~50 of them are actual users. That's roughly 33%.

Why It Surprises People: The Base Rate Fallacy

Most people focus on the "99% accurate" figure and ignore the base rate — the fact that users are extremely rare (0.5%).

The key insight: false positives come from a much larger pool. Even though the false positive rate is small (1%), it applies to the 99.5% of people who don't use the drug. So false positives (99.5 people) actually outnumber true positives (49.5 people) two to one.

Our intuition treats "the test is 99% accurate" as if it means "a positive result is 99% correct" — but those are entirely different quantities. The test's accuracy in detecting users says nothing directly about how much to trust a positive result, which depends critically on how rare the condition was to begin with.

Practical implication: In mass screening for rare conditions (drugs, diseases, terrorism watchlists), a positive result alone is often weak evidence — confirmatory testing is essential.