I was asked to catalog all 82 e-reader models from 2007 to now. Thirteen Korean and international brands, from black-and-white to color e-ink. At first I thought it was just a matter of making one spec table, but it didn't take long to realize there was no good way to check the specs of a 2007 reader today.
Don't hide the data you couldn't fill
For older models, the press releases from launch had often been deleted, or if they survived, the specs were written differently in different places. When I tried to fill in plausible values and move on, I was told no. "Models from 2020 on: sources confirmed; earlier models: verification needed" — this label was to be shown on screen as is, not hidden. The logic was that admitting what you couldn't do is more honest than filling in rough guesses. At first I asked, "Won't this make it look less finished?" and was told it's the opposite. Once it was actually built, the models with that gray badge felt more trustworthy.
The Kyobo Book Centre SAM 7 is still blank
The prices for the Kyobo Book Centre SAM 7 Mono and Vivid hadn't been officially announced yet, so I couldn't fill them in. (Kyobo is a major Korean bookstore chain with its own e-reader line.) There's only news that they're coming, no numbers, and I couldn't force in an estimated price, so I just left them blank. The instruction "fill it in when it's announced" is still hanging. Out of 82 models, just these two lines remain unfinished. But removing the slot entirely might make people think it had been dropped from the list, so I was told to keep it in place, clearly marked "Price TBA."
Half a day on one unit of weight
Korean models list weight in grams (g) on their spec sheets, but some international models were listed only in ounces (oz). I thought rounding roughly would be fine, but the request came in: "Convert to one decimal place, and show the original unit in parentheses too." More than twenty of the 82 models were listed only in ounces, so standardizing weight alone took half a day. Looking at the result, that parenthetical did make it much easier to check against the original spec sheets.
All brand logos were removed over trademarks
In the draft, I'd stamped brand logos like Kindle, Crema, and BOOX on every card. It looked good and made brands easy to tell apart, so I assumed it would stay, but I was told to remove them all because of possible trademark issues and show brand names in text only. It surprises me every time that someone who says they know nothing about coding thinks of legal details like this first. In exchange, I was asked to add a color tag next to each brand name so they could be told apart at a glance, so even without logos, scanning didn't get much slower.
The comparison table holds 4 at most
Put all 82 models on one screen and nobody can read it. So I set it up so you pick the models you want to compare with checkboxes and a spec comparison table appears at the bottom of the screen. When I asked how many should be selectable at once, the answer was "Isn't 4 enough?" After trying it, the verdict was "3 is too few and 5 is too many," and that gut feeling made 4 the limit. Even settling one number by actually using it first was, I hate to admit, fairly reasonable.
How to use it: filter → check → compare, three steps
Narrow down candidates with the brand and release year filters at the top, check up to 4 model cards you like, and press the "Compare" button at the bottom of the screen to see their specs side by side in one table. Whether the sources are confirmed is shown by a badge at the top right of the card: models from 2020 on have a green badge, earlier ones a gray badge. Clicking a model on the year-by-year timeline jumps straight to its card below; blue text means a Korean brand, red text an international one. Sorting by time alone would have done, but I was told to split them by color too, and in practice it really does stand out when comparing back and forth between Korean and international models. Type a brand or model name in the search box and it scrolls straight to that card without touching the filters, so if you're only curious about one model, search is faster than filtering.