Subject: Heating repair—request for confirmed completion date

Dear [Landlord’s Name],

I’m following up on the heating issue I first reported on [date]. It has now been three weeks, and the heating is still not working. This is affecting my ability to live comfortably in the property and needs to be resolved promptly.

Please confirm by [date] when a qualified technician will attend and when the repair is expected to be completed. If there is a delay with parts or scheduling, please let me know what temporary heating arrangements you can provide in the meantime.

I’d like to resolve this cooperatively. My priority is simply to have reliable heating restored, and I’m happy to coordinate access to help get the work done.

Thank you for your prompt attention. I look forward to receiving a confirmed plan.

Kind regards, [Your Name] [Property Address]

Subject: Heating repair at [property address] — outstanding for three weeks

Dear [Landlord's name],

I hope you're well. I'm writing to follow up on the heating at [property address/unit], which I first reported on [date]. It's now been three weeks without working heating, and despite [my messages on X and Y dates / my calls on X and Y], the repair hasn't yet been carried out or scheduled.

With the cold weather, this is becoming a genuine problem — [briefly note the impact, e.g., the flat is very difficult to keep warm and I'm concerned about the effect on my health]. I'd ask that this now be treated as urgent.

Could you please confirm by [date — e.g., within five working days] when a contractor will attend? I'm happy to be flexible with access and can make myself available [on these days / at short notice] if that helps.

I'd much rather resolve this directly with you than take it any further, and I'm confident we can get it sorted quickly. I'd really appreciate a reply confirming the plan, even if the appointment itself is a few days away.

Thank you for your attention — I look forward to hearing from you.

Kind regards, [Your name] [Phone number]

A few quick tips:

• Fill in the specifics (dates you reported it, any responses received) — a clear timeline makes it harder to brush off.

• Give a concrete deadline rather than "as soon as possible"; a week is usually seen as reasonable.

• Keep everything in writing from here on. This email becomes your record if you do ever need to escalate to a tenants' advice service or local authority — hopefully you won't.

The tone stays cooperative throughout, but the deadline and the one line about "taking it further" signal you're serious without starting a fight.

import { useEffect, useState } from "react";

/**
 * Returns the value after it has stopped changing for `delayMs`.
 * Cancels pending updates when the value/delay changes or on unmount.
 */
export function useDebouncedValue<T>(value: T, delayMs: number): T {
  const [debouncedValue, setDebouncedValue] = useState<T>(() => value);

  useEffect(() => {
    const timeoutId = setTimeout(() => {
      // The wrapper also supports function-valued T.
      setDebouncedValue(() => value);
    }, delayMs);

    return () => {
      clearTimeout(timeoutId);
    };
  }, [value, delayMs]);

  return debouncedValue;
}

Usage:

import { useState } from "react";
import { useDebouncedValue } from "./useDebouncedValue";

export function SearchInput() {
  const [query, setQuery] = useState("");
  const debouncedQuery = useDebouncedValue(query, 300);

  return (
    <div>
      <input
        value={query}
        onChange={(event) => setQuery(event.target.value)}
        placeholder="Search…"
      />
      <p>Debounced query: {debouncedQuery}</p>
    </div>
  );
}

The initial value is returned immediately. Subsequent changes are delayed until delayMs has elapsed without another change; effect cleanup cancels the pending timer on unmount.

useDebouncedValue

import { useEffect, useState } from 'react';

function useDebouncedValue<T>(value: T, delay: number = 500): T {
  const [debouncedValue, setDebouncedValue] = useState<T>(value);

  useEffect(() => {
    const timer = setTimeout(() => {
      setDebouncedValue(value);
    }, delay);

    // Cleanup runs when value/delay changes AND on unmount
    return () => {
      clearTimeout(timer);
    };
  }, [value, delay]);

  return debouncedValue;
}

export default useDebouncedValue;

How it works

• Generic <T> — the hook preserves whatever type you pass in (string, number, a custom object, etc.), so the returned value is fully typed.

• Debouncing — every time value changes, the effect re-runs. The cleanup function clears the pending timer before a new one starts, so the state only updates once the value has been stable for delay milliseconds.

• Unmount safety — the same cleanup function runs when the component unmounts, cancelling any pending timeout. This prevents a setState call on an unmounted component (and the associated memory leak / React warning).

Usage example

import { useEffect, useState } from 'react';
import useDebouncedValue from './useDebouncedValue';

function SearchPage() {
  const [query, setQuery] = useState<string>('');
  const debouncedQuery = useDebouncedValue(query, 300);

  useEffect(() => {
    if (!debouncedQuery) return;

    // Only fires 300ms after the user stops typing
    fetch(`/api/search?q=${encodeURIComponent(debouncedQuery)}`)
      .then((res) => res.json())
      .then(console.log);
  }, [debouncedQuery]);

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

The input stays responsive on every keystroke, while the expensive operation (the API call) only runs after typing pauses.

About 33.2%—assuming “99% accurate” means both 99% sensitivity and 99% specificity. That is, the test detects 99% of users and correctly gives a negative result to 99% of non-users.

Imagine testing 20,000 people:

Of the 298 positive results, only 99 are from actual users:

[ P(\text{uses drug}\mid\text{positive})=\frac{99}{99+199}\approx \boxed{33.2\%}. ]

Why it surprises people: We tend to confuse “the test is positive in 99% of users” with “99% of people who test positive are users.” Those are different probabilities. Because non-users vastly outnumber users, even a 1% false-positive rate produces more false positives than true positives. Ignoring this imbalance is called base-rate neglect.

If “99% accurate” refers only to overall accuracy, rather than sensitivity and specificity separately, there isn’t enough information to calculate the answer.

The answer: about 33%

Despite the positive result, there's only roughly a 1-in-3 chance the person actually uses the drug.

Working through it

Assume "99% accurate" means the test catches 99% of users (sensitivity) and correctly clears 99% of non-users (specificity). Now imagine testing 100,000 people:

Total positive tests: 495 + 995 = 1,490

$$P(\text{user} \mid \text{positive}) = \frac{495}{1{,}490} \approx 33\%$$

Same result via Bayes' theorem:

$$P(U \mid +) = \frac{0.99 \times 0.005}{0.99 \times 0.005 + 0.01 \times 0.995} = \frac{0.00495}{0.0149} \approx 0.33$$

Why it surprises people

1. Base rate neglect. People anchor on the 99% accuracy and ignore the 0.5% prevalence. But because users are so rare, the false positives (995) actually outnumber the true positives (495) two to one. A small error rate applied to a huge pool of non-users swamps the correct results from the tiny pool of users.

2. Confusing two different probabilities. People mix up:

• P(positive test | user) = 99% — what "accurate" means

• P(user | positive test) = 33% — what you actually want to know

These feel identical but are wildly different when the condition is rare.

3. Intuition treats the test as the whole story. The test result is evidence that updates a prior belief, not a verdict. A positive result on a rare condition is weak evidence; a positive result on a common one would be much stronger.

The practical upshot

This is exactly why confirmatory testing exists. If the same person takes a second, independent test and it's positive again, the probability jumps to about 98% — because you're now testing someone with a 33% prior, not 0.5%. It's also why screening entire populations for rare diseases inevitably generates many false alarms.