Here is a polite but firm template you can copy, paste and customize. It's designed to get the repair done while keeping the relationship professional:

Subject: Follow-up: Heating repair request at [Your Address / Apt #] - Outstanding 3 weeks

Dear [Mr./Ms. Last Name],

I hope you are well. I am writing to follow up regarding the heating issue at [Your Full Address, Apt #].

I first reported that the heating was not working on [Date of first report - e.g., January 20, 2026], and have followed up on [Date of 2nd contact] and [Date of 3rd contact]. As of today, it has now been three weeks without functioning heat.

With the cold weather, this is becoming difficult to live with, and I would like to get this resolved as soon as possible.

Could you please arrange for a qualified technician to repair the heating by [Give a reasonable, specific deadline - e.g., Friday, February 13]? I am available for access on [List 2-3 dates/times you are available, e.g., weekday afternoons after 3pm and anytime Thursday/Friday] but I am happy to adjust if another time works better for your contractor.

Please could you let me know in writing what the plan and timeline for the repair will be? I appreciate that repairs can take time to coordinate, and I want to work with you to get this sorted.

Thank you very much for your help with this. I look forward to your reply.

Kind regards, [Your Full Name] [Your Address / Apt #] [Your Phone Number]

Tips to make it effective:

1. Keep a record: Send it by email and keep a copy. Note the dates you reported it previously.

2. Give a clear deadline: 3-5 business days is usually considered reasonable and firm.

3. Offer access: Landlords often delay because they say they couldn't get in. Offering times removes that barrier.

4. Stay solution-focused: The tone here is "we both want this fixed" rather than accusatory.

If you get no response after your deadline, you can add this one sentence in a follow-up: "As heating is considered an essential service, I want to resolve this directly with you before I look into other options for advice and support."

Want me to tailor this with your exact dates, address, and landlord's name?

This hook debounces any value and automatically cancels the pending timeout on unmount via useEffect cleanup:

import { useState, useEffect } from 'react';

export function useDebounce<T>(value: T, delay: number): T {
  const [debouncedValue, setDebouncedValue] = useState<T>(value);

  useEffect(() => {
    // Set a timer to update the debounced value after the delay
    const timerId = setTimeout(() => {
      setDebouncedValue(value);
    }, delay);

    // Cleanup: cancels pending update on value/delay change or on unmount
    return () => {
      clearTimeout(timerId);
    };
  }, [value, delay]);

  return debouncedValue;
}

Usage

import { useState } from 'react';
import { useDebounce } from './useDebounce';

function SearchInput() {
  const [query, setQuery] = useState<string>('');
  const debouncedQuery = useDebounce<string>(query, 500);

  // This effect only fires after user stops typing for 500ms
  useEffect(() => {
    if (debouncedQuery) {
      // fetchResults(debouncedQuery)
    }
  }, [debouncedQuery]);

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

The return () => clearTimeout(timerId) inside useEffect guarantees no state update is attempted after the component unmounts or if value changes before delay expires.

This is a classic Bayes' Theorem problem. The answer is about 33%, not 99%.

Let:

• D = you use the drug

• + = you test positive

We are given:

• P(D) = 0.5% = 0.005 -> P(not D) = 0.995

• P(+|D) = 99% = 0.99 : If you do use, you test positive 99% of the time (sensitivity)

• P(not +|not D) = 99% = 0.99 : If you don't use, you test negative 99% of the time (specificity)

Which means: P(+|not D) = 1% = 0.01 : False positive rate

You want P(D|+) - if you tested positive, what is the chance you actually use?

People confuse P(+|D) with P(D|+). The test is 99% accurate when we know you use. We don't know the reverse.

By Bayes' Theorem:

$$P(D|+) = \frac{P(+|D) \cdot P(D)}{P(+)}$$

$$P(+) = P(+|D)P(D) + P(+|not D)P(not D)$$

$$P(+) = 0.99 \times 0.005 + 0.01 \times 0.995$$ $$P(+) = 0.00495 + 0.00995 = 0.0149$$

So:

$$P(D|+) = \frac{0.00495}{0.0149} \approx 0.332 = 33.2\%$$

If you test positive, there's only about a 1 in 3 chance you actually use the drug.

Why it's so surprising: The Base Rate Fallacy

Think in natural frequencies instead of percentages. Imagine 10,000 people:

1. 50 are users (0.5%): 99% test positive -> ~49 true positives

2. 9,950 are non-users: 1% test positive -> ~100 false positives

Total positives: 49 + 100 = 149

True users among positives: 49 / 149 ≈ 33%

Our brain does two things wrong:

1. We ignore the base rate. 0.5% is extremely rare. There are ~200x more non-users than users. Even a tiny 1% error rate on that huge group of non-users creates twice as many false positives as there are real positives.

2. We confuse the conditional. We hear "99% accurate" and think it means P(user | positive) = 99%. It doesn't. It means P(positive | user) = 99%.

This is why doctors don't panic at a single positive result for a rare disease, and why a second independent test is so powerful. If you tested positive twice in a row, the updated probability jumps to ~98%.