I don’t like Fran Blanche. For someone who’s a senior member of the Institute of Electrical and Electronics Engineers (IEEE), her videos are incredibly shallow.

Five years ago, Fran made a video on “Just How Terrible Is A NEW Portable Radio???”, in which she presented a DSP radio based on KT0936MB9. Its schematic was practically identical to the one used in the Philips radios I discussed in “Fun and pain with a teardown—analogic radios are defunct” and only slightly different to the one from the Aiwa R-190 described in Romanian in another post. Aiwa has a fake bass boost that I suspect is due to some passive circuit that I couldn’t be bothered to identify, but it seems to be cutting the treble, too.

Some two weeks ago, Fran posted another video: “Can A Cheap Radio Be Made Better?” It was about the same device!

The major problem was that the audio potentiometer in her cheap radio was starting to fall apart. In the process of cleaning and fixing things, she introduced a slight modification around the audio potentiometer. On the one hand, this modification extended the usable range of the potentiometer (the sound gets distorted towards the maximum, which limits the usable range), and on the other hand, it sort of faked a bass boost by cutting the high frequencies at high volumes.

Enter Gemini

I wasn’t happy with her idea, so I had a quick chat with Gemini, starting with this prompt:

Replacing in an audio circuit a potentiometer with the shown simple circuit affects the frequency response, especially at higher volumes. Can you draw a chart of the frequency response relative to the position of the potentiometer?

To cut a long story short, with Fran’s circuit, the frequency response is horrendous:

This is not how you fake a bass boost, no matter how cheap you want it to be!

Gemini suggested the following improvement:

The resulting frequency response becomes much more interesting:

However, I should probably have opted for the first value proposed by Gemini, 1 nF. With 2.2 nF, at 100% the high frequencies aren’t cut anymore, quite the opposite! The “bright capacitor” is too effective.

Then, Gemini offered to chart the effects of changing the value of the 10 kΩ resistor:

Quite interesting, I’d say.

Food for thought

Food for thought: why is “kΩ” using a minuscule k, whereas “MΩ” is using a majuscule M, despite both increasing the value of the unit? When “mA” is using a minuscule m, the prefix is diminishing the value of the unit.

Blame the International System of Units (SI). Or IEC 80000-13.

According to it, only the units that are named after a person are capitalized: Ohm, Volt, Ampere, Kelvin, Celsius, Fahrenheit. Multipliers like k (kilo) and demultipliers like m (milli), c (centi), and d (deci) are not.

Kilo must be k; otherwise, it would clash with Kelvin.

However, M (Mega), G (Giga), and T (Tera) are capitalized primarily to avoid confusion with other units; specifically, Mega and mili can’t have the same capitalization.

Things become kinkier when you get to binary data.

The KiB (kibibyte) has been defined by the binary prefixes standard IEC 60027-2 to mean 210 = 1,024 bytes. And not only have we (capitalized) KiB and kB/KB, but the confusion is ginormous.

Historically, in computing, “kilo” was used to mean 1,024 (210), and “mega” was used to mean 1,048,576 (220). JEDEC/Windows popularized them written in capitals: KB, MB.

Purists insisted that “k” was supposed to mean 1,000 (103), so IEC 60027-2 stepped in and standardized this convention:

  • kB (kilobyte): 103 = 1,000 bytes. In practice, KB (technically incorrect) is often used for this too.
  • KiB (Kibibyte): 210 = 1,024 bytes.

But most people still use KB to mean KiB! Some file managers also display MB or GB when they mean MiB or GiB. And this creates a double confusion:

  • Is KB supposed to mean 1,024 (JEDEC), as a wrong unit for KiB?
  • Is KB supposed to mean 1,000, as a wrong capitalization of kB?

I asked Kimi the following question:

Can you make a list of which of these file managers display the correct prefixes and which use KB/MB/GB instead of KiB/MiB/GiB?

  • File Explorer in Win7
  • File Explorer in Win10
  • File Explorer in Win11
  • Dolphin (KDE)
  • Caja/Nautilus (MATE)
  • Files/Nautilus (GNOME 3+)
  • Thunar (XFCE)
  • Nemo (Cinnamon)
  • PCManFM/PCManFM-Qt

If I’m not wrong, at least one of them can be customized as to which units to display.

It answered:

Here’s the breakdown for your list of file managers:

File ManagerDefault DisplayNotes
File Explorer (Win7/10/11)KB, MB, GB (but meaning 1024 bytes)Uses JEDEC standard—displays KB/MB/GB labels but calculates using base-2 (1024). No option to change.
Dolphin (KDE)KiB, MiB, GiBUses proper IEC binary prefixes. Was customizable in KDE Plasma 4 via BinaryUnitDialect in kdeglobals (0=IEC, 1=JEDEC, 2=metric), but this setting appears non-functional in Plasma 5+—now locked to IEC units.
Caja (MATE)kB, MB, GB (base-10)Follows the GNOME/Ubuntu “Units Policy”—uses SI decimal prefixes.
Nautilus/Files (GNOME 3+)kB, MB, GB (base-10)Same as above—switched to SI decimal units in Ubuntu 10.10+ / GNOME 3.x era.
Thunar (XFCE)ConfigurableThis is the customizable one! Defaults to KB/MB/GB (base-10), but has a checkbox in Preferences → View: “Show file size in binary format” which switches to KiB/MiB/GiB.
Nemo (Cinnamon)kB, MB, GB (base-10)Similar to GNOME/MATE—uses SI units.
PCManFM/PCManFM-QtKiB, MiB, GiBUses IEC binary prefixes by default.

The winner for customization: Thunar is indeed the one you can toggle. It offers the best of both worlds with a simple preference checkbox.

The irony: Dolphin is the only one that used to be highly customizable (three different dialects!) but regressed—users now complain about being locked into KiB/MiB/GiB when they want MB/GB for cross-platform consistency.

The Windows situation: Explorer is arguably the worst offender—not because it uses base-2 (which is defensible for file systems), but because it uses the wrong labels for it (KB instead of KiB), perpetuating the confusion you mentioned. Microsoft has stubbornly maintained this since the JEDEC 100B.01 standard, despite the IEC 60027-2 standard existing since 1999.

So the landscape is split: KDE/Qt file managers (Dolphin, PCManFM) tend toward correct IEC binary prefixes, while GNOME/GTK file managers (Nautilus, Caja, Nemo) and Windows use SI labels—but Windows calculates in binary while GNOME calculates in decimal. Thunar sits in the middle, letting you choose your poison.

But wait, there is more! How about networks speeds? Or modem speeds, historically speaking.

Question to Kimi:

But is the minuscule “kB” still supposed to represent something? Are “kbps” (bits) or “kB/s” (bytes) legit?

And its answer:

Excellent follow-up! You’re touching on the other great SI confusion: bits vs. bytes, and yes, the minuscule “k” has different meanings depending on whether you’re talking about bytes or bits.

The “k” in kB vs kb

NotationMeaningStandard
kB (kilobyte)1,000 bytesSI metric (IEC 80000-13)
KB (kilobyte)1,024 bytesJEDEC / Windows convention
KiB (kibibyte)1,024 bytesIEC 60027-2 binary
kb (kilobit)1,000 bitsSI metric (always)
kbit (kilobit)1,000 bitsSI preferred
Kibit (kibibit)1,024 bitsIEC binary

Is “kB” still legit?

Yes, kB (lowercase k) is the correct SI form for 1,000 bytes. The SI system dictates:

  • Lowercase k = kilo (10³) — applies to all units
  • Uppercase K = Kelvin — only for temperature

So strictly speaking:

  • kB = 1,000 bytes (correct SI)
  • KB = technically wrong for SI, but means 1,024 bytes in computing JEDEC tradition

kbps vs kB/s — The Bit/Byte Trap

This is where marketing loves to mess with you:

NotationMeaningCommon Usage
kbps or kb/skilobits per second (1,000 bits/s)Network speeds (ISP marketing)
kB/s or kBpskilobytes per second (1,000 bytes/s)File transfers (rare)
KB/skilobytes per second (1,024 bytes/s)Windows file copies
KiB/skibibytes per second (1,024 bytes/s)Proper technical usage

The gotcha: ISPs advertise speeds in Mbps (megabits) because the number is 8× bigger than MB/s. A “100 Mbps” connection sounds faster than “12.5 MB/s” — even though they’re the same thing.

Your Specific Question

Are “kbps” (bits) or “kB/s” (bytes) legit?

Yes, both are legit — but context matters:

  • kbps = always bits (networking convention)
  • kB/s = bytes with SI decimal prefix (1,000 bytes/s)
  • KB/s = bytes with binary/JEDEC prefix (1,024 bytes/s) — Windows style

The lowercase k in kbps is correct SI. The lowercase k in kB/s is also correct SI for 1,000. The confusion arises because:

  1. Computing co-opted K (uppercase) to mean 1,024
  2. Networking stuck with SI k = 1,000 for bits
  3. Nobody uses Kibps for 1,024 bits/s (though they should)

So when you see “kbps,” it’s unambiguously 1,000 bits per second. When you see “kB/s,” it should be 1,000 bytes per second, but many people use it interchangeably with “KB/s” (1,024). The only way to be unambiguous is to use KiB/s for 1,024 or stick to context.

The real crime? Storage manufacturers use GB (1,000,000,000 bytes) to make drives look bigger, while operating systems report GiB (1,073,741,824 bytes) but label it as GB — making your “1 TB” drive show up as ~931 “GB” in Windows.

C’est à perdre la raison…

All this confusion could have been avoided if and only if purists had chosen to go fuck themselves and therefore accept that:

  • In computing, “K” is meant to be 1,024 (base-2) not 1,000 (base-10). JEDEC’s rules rulz.
  • In computing, “K” is meant to be capitalized because there’s no possible clash with Kelvin, therefore ensuring the consistency with other prefixes (M, G, T).