9. The Score Is Someone Else's Weights

You have a reading-needs record, and you have spent several lessons learning what each column of a comparison table actually measures. This lesson is the buying decision. The question is simple: how do you turn a table of fifty-seven devices into a short list you can defend, with a real cost attached? The answer starts with a warning. The total score on a comparison page is a set of weights that someone else chose. Your own requirements come first, and they work as filters, not as points.

Everything here uses OCDevel's comparison table at ocdevel dot com slash ereaders, as published on the thirtieth of October, 2026.

Filters before scores

Start by writing down your deal-breakers. A deal-breaker is something where "almost" is the same as "no." A weighted score cannot respect that, because it adds up points. A device can be excellent on five things, collect a high total, and still lack the one feature you cannot live without. The strong extras cover for the missing essential. So you eliminate first and score second.

The deal-breakers people most often have are these. Warm front light, if you read in bed. A weight ceiling. An operating system open enough to run a library app or another third-party app. Physical page-turn buttons. Water resistance. The table lets you filter on most of these, so apply them as filters before you read a single total. Filters combine, so you can ask for warm light and buttons and a water rating at the same time, and the list that remains is your real field.

What the total counts

Now look at what the total is. OCDevel describes it as a score from zero to one hundred for ordinary book reading. Several things count directly: front light, text clarity, weight, document transfer, price, battery and physical controls. Several things do not add points, though they stay visible and filterable: RAM, stylus, audio, colour and screen size. So a note-taker with a superb pen gets no credit on the total for that pen. The total is not a verdict on the device. It is a verdict on the device as a book reader, by one publisher's standards.

A few details are worth knowing. Price does not enter as a plain number. It goes in through a harmonic mean of quality and affordability. A harmonic mean is pulled down hard by whichever of the two is weak, so a cheap device of poor quality and an excellent device that is very expensive both get held back. Weight is a straight ranking against the rest of the table: the lightest device scores ten, the heaviest zero, and the others fall in between. That ignores screen size, so a thirteen-inch note-taker is judged against a pocket reader. Storage reaches its top score at sixteen gigabytes, so more adds nothing to the total, even though a large PDF library would still want it. And brand is an attributed editorial judgment about reputation. It is not a measured property of the hardware.

Handing out your own weights

Here is how weights change an outcome. This is an illustration of the method, not OCDevel's own scoring. Take two devices, the Kobo Libra Colour and the Boox Go 7. Give each a score out of ten on three things. The Libra Colour gets seven for contrast, nine for ecosystem and eight and a half for colour. The Go 7 gets nine and a half for contrast, seven and a half for ecosystem, and zero for colour.

Now weight them as a general reader might: forty percent contrast, thirty ecosystem, thirty colour. The Libra Colour scores eight and five hundredths. The Go 7 scores six and five hundredths. The Libra wins comfortably.

Now suppose you read only black and white and do not care about colour. Weight contrast at sixty, ecosystem at thirty-five, colour at five. The Go 7 now scores eight point three two five, and the Libra Colour seven point seven seven five. The order reverses. Hold the contrast-to-ecosystem ratio fixed, and the tipping point sits at about ten point eight percent weight on colour. Below that, the Go 7 leads. Above it, the Libra.

So here is the method for you. Hand out a hundred points across the things you care about, according to what actually bothers you in your current setup. Then take your biggest weight and move it five to fifteen points up and down. If the same device keeps winning, the ranking is sturdy. If the winner flips, the choice depends on a judgment you should think harder about, and that is where to spend your testing time.

Three short lists

The sample readers from earlier lessons make this concrete. These lists use only devices on the table, and being on the table is not the same as being a pick.

The novel reader borrows about eighty books a year, reads in bed and on a beach, and wanted sharp black-and-white text. The filters are an IPX8 rating, warm light, and preferably page buttons. Four devices have that rating and warm light. The Kindle Colorsoft and the Kobo Clara Colour add a colour screen but no buttons, which leaves two candidates worth this reader's time. The Kobo Libra Colour is OCDevel's top non-Kindle pick, with native OverDrive lending, page buttons, IPX8 and an asymmetric grip. The Kindle Paperwhite, in its Signature form, is OCDevel's value default, also IPX8, but with no buttons. The tradeoff is that the Libra's colour layer halves effective pixel density in coloured areas, and this reader wanted monochrome sharpness. Whether Kobo's library lending works with this reader's country and library is still unconfirmed, and that has to be verified before anything else.

The technical reader needs room for two-column papers and a large local PDF library. The Boox Go 10.3 Gen II Lumi is OCDevel's medium-size pick: 300 pixels per inch, Android 15, a front light. The Kindle Scribe is an eleven-inch option. The Boox Note Max, at 13.3 inches, has no front light, which suits a desk and rules it out for bed.

The note-writing reader looks at the Kindle Scribe, the Boox Note Air6 C with its Pen3 in the box, and the reMarkable Paper Pro. For the Paper Pro, add the Connect subscription to the cost. Without it, the cloud keeps only documents opened or synced in the previous fifty days. The documents stay on the device itself, but your copies elsewhere do not.

What it really costs

Sticker price is one line. As of October 2026, the Paperwhite is $199.99 for sixteen gigabytes with lock-screen ads, or $249.99 for the thirty-two gigabyte Signature Edition without ads. A fabric folio case is $36.99. The Colorsoft is $289.99, or $319.99 for the Signature. Cases for it run $36.99 to $49.99. On Kindle base models, ad-free costs twenty dollars more. The Kindle Scribe without a front light is $429.99. Boox gives ten gigabytes of free cloud storage and no required subscription. Standard shipping inside the United States is generally free on direct orders that meet the brand's purchase threshold.

For every other price, use the live price on the table. Sales tax is not covered by this evidence, so check it at checkout. Add it up as device, ads decision, case, pen if it is not included, subscription for a year, and tax.

Warranty and returns

Amazon offers a thirty-day return and a one-year warranty, with chat support at any hour. Kobo gives one year in the United States, round-the-clock chat, and a repair partnership with iFixit. Boox and reMarkable both offer one year. The return window matters most. It is the period in which you can test the device against your hardest document, check the front light for uneven glow at the edge, and watch overnight battery drain. A test you cannot run before the window closes is not a test.

Sometimes nothing passes the filters. Then keeping your current device, or buying nothing, is the correct result.

This week's changes

The new Kindle Paperwhite and Colorsoft went on sale on the twenty-eighth of October. Amazon's battery claim assumes thirty minutes of reading a day, Wi-Fi off, and the light at level thirteen. In one hands-on test, mixed reading with varied brightness and downloads drained the Colorsoft about thirty-one percent in a week. If you read an hour or more a day with wireless on, expect much less than the headline weeks. The Kindle Click page-turn remote sold out, and new orders are quoted for mid-November or later.

If you own a Kobo, software release 6.0.279225 came out on the twenty-eighth of October. It fixes data-loss bugs that affected devices on two earlier versions, including lost reading stats, saved words and sideloaded books. Update, then open your library and check that your own files are all there. Kobo has also renamed its OverDrive menu entries to Libby, and Send-to-Kobo now works with the CloudLibrary service. Kobo's borrowing from several library cards at once has been announced but has still not shipped.

The Boox Picco appeared in sporadic stock on Amazon in the United States in late October, then sold out. It still needs the BOOX Tiles app to load your own files. And Kaleido 4, E Ink's new colour panel, is still confirmed in only one device, the Hanvon M10C II, limited to China. So there is nothing yet to test.

Your next step

The filters on the comparison page at ocdevel dot com slash ereaders are your deal-breakers in practice. One thing to know: the table earns a commission on its links, and a commission never buys a pick.

Today, write a dated short list. Put down your deal-breakers, your hundred points of weights, a budget, and the full cost of each device on it. Then name the one title or document you cannot live without, and plan to test it end to end inside the return window. That test is next.