Corrections
Every change to a published number, newest first, with what changed and why.
When a published number changes, it goes here: the date, what changed, and which cards and pages it touched.
The score now reads 22 nutrients
- What you would have seen
- The score read 18 nutrients. Vitamin D, choline and the omega-3s were listed on food and nutrient pages but counted for nothing, so a food rich in them got no credit for it.
- What you see now
- Vitamin D, choline, the plant omega-3 ALA and the fish omega-3s EPA and DHA count too (method v0.4). Fatty fish, eggs, walnuts, flax and chia moved up the Everyday ranking: sardines from 30th to 10th, walnuts from 186th to 138th. Greens, lentils and sesame seeds, which carry little of the four, moved down: boiled lentils from 27th to 42nd. Reviewed foods missing a value for one of the four were filled first from the national food tables, where a table lists one. The calcium card did not change.
Lactose was read from the wrong line of the USDA tables
- What you would have seen
- Every lactose figure taken from USDA FoodData Central was the food’s leucine, an amino acid: the data read nutrient 1213 (leucine) instead of 1013 (lactose). Grilled steak showed 2.1 g of lactose per 100 g and whole milk 0.33 g.
- What you see now
- Lactose comes from the right line: whole milk 4.81 g, 2% milk 5.01 g and skim milk 5.09 g per 100 g, and none in meat or fish. Leucine is listed under its own name. The Canadian and UK lactose figures were always right. Foods not reviewed yet show no USDA lactose until their next import. No score uses lactose, so no score or rating changed.
Pages affected: Foods
Contaminants with no safe level lost their limit labels
- What you would have seen
- For contaminants that have no intake known to be safe (inorganic arsenic, acrylamide, aflatoxins, furan, lead and the like), some foods read “Can push past the limit”, “Low at ordinary amounts” or “Under the limit at ordinary amounts”. There is no limit for those, so the labels contradicted the pages around them.
- What you see now
- They read “Can double average intake” or “Doesn’t double average intake”, from how much a daily serving adds to what an average adult already takes in. A row with no such share reads “Rated by margin of exposure”. Only the labels changed; no rating moved.
The pages gave different food counts
- What you would have seen
- Pages disagreed on how many foods are reviewed. How to read this site and Method counted the calcium study’s fresh-cheese entries and its reference milk as foods (62 foods, 102 preparations); the Foods list and the verdicts did not (60 and 97). The home page said 88 foods and the hazards page “87 of the 88”.
- What you see now
- Every page counts the foods you can look up, by one rule. On 20 October that gave 86 foods and 139 preparations; foods have been added since, and the pages now read 159 foods and 229 preparations. The counts are read from the database, not typed, so every page shows the same current figure.
Pages affected: Food safety · Method · How to read this
Everyday ranking and Better options changed (method v0.3)
- What you would have seen
- Everyday compared nutrients per 100 calories and counted calcium at its listed amount. Better options compared a food with any other, so cola, coffee and ice cream were all offered raw watercress, among other mismatches, and foods with more oxalate than their own calcium can bind were suggested as swaps.
- What you see now
- Everyday compares what one serving gives, counting at most 100 calories of it. Where the calcium card measured that less calcium gets in than from milk, calcium counts for less. Foods high in oxalate are no longer suggested as swaps. Better options compares a food only with foods eaten the same way: drinks with drinks, desserts with desserts, starchy sides with starchy sides. The lose-weight and gain-weight scores kept their recipes, but the calcium change reaches them too: raw spinach, cooked soybeans and boiled green beans score slightly lower, and for the gain-weight goal raw spinach moved below cooked soybeans and raw watercress. The calcium card did not change.
Grams to the limit now count from the room left
- What you would have seen
- For cadmium, dioxins, methylmercury, nickel, 3-MCPD and deoxynivalenol, the grams that reach the limit were counted from zero, as if the food were all you ate: raw spinach, cadmium, 1,623 g a week.
- What you see now
- They count from the room left after an average adult’s intake, so the figures are smaller: raw spinach 474 g, whole flaxseed 476 g → 139 g, boiled pasta 5,070 g → 1,480 g (dioxins excepted: an average adult’s intake is already over the limit, so they still count from zero). Three cadmium ratings moved with them, from “Low at ordinary amounts” to “Can push past the limit”: raw spinach, whole flaxseed and boiled pasta. The cadmium measured in those foods did not change, and no other rating moved. The fish mercury, spinach and cadmium receipt cards were written before this and still print the figures from zero; each now says so.
Pages affected: Foods · Food safety · Check your week
Method v0.2 changed some scores
- What you would have seen
- Niacin was scored as plain niacin, not as niacin equivalents, the form its reference intake is set in. Egg, farmed salmon, ground beef, dry lentils, rolled oats and almonds were scored with nutrients missing that USDA’s Standard Reference table does list for the same food. Pumpkin seeds carried a vitamin E value that belonged to a different preparation.
- What you see now
- Niacin is scored as niacin equivalents. Those six foods gained the missing values, which now count. A food’s density is an average over the nutrients it has, so several scores fell: for the lose-weight goal, farmed salmon went from 6th to 9th of the 20 comparison foods. Pumpkin seeds lost the vitamin E value that was not theirs. Cadmium shares are counted against the European adult median intake (1.77 µg per kg of body weight a week). The calcium card did not change.