图形学与人机交互2021harold.thimbleby.net
不要使用七段数码管（2011）
Don't use 7-segment displays (2011)
入门 · 约 25 分钟读完前置知识：基础界面设计、人机交互常识、数字显示基础
导读
这篇论文（Harold Thimbleby，2011）批评在界面设计中滥用七段数码管：它只适合显示数字，用于字母、状态或安全关键信息时容易产生歧义和误读。作者结合案例主张，改用高分辨率显示往往能提升可读性、安全性和灵活性，设计者不应被七段管的低成本惯性和定制面板方案束缚。
HN 关注点：HN 读者既用公交站屏、酒厂误配置等实例支持“重要场合别用七段管”，也从硬件成本与开发复杂度角度质疑作者对高分辨率方案过于乐观，讨论焦点是可读性/安全性与成本/复杂度如何权衡。
HN 评论精选
这几条评论里：质疑 2 补充 1 亲历 1
质疑公交站从低分辨率 LED 点阵换成高分辨率 LCD 后反而更难读，说明升级不一定改善信息传达。
在牛津市中心（可能其他地方也一样），公交站的实时信息屏被换掉了，这些屏显示下一班车何时到、时间和日期等。以前用的是琥珀色 LED 点阵：对比度高、易读、分辨率低。现在换成了彩色 LCD 面板：高分辨率、高科技，可能成本也高。它们可以显示公交公司标志、平滑滚动文字和任意图片——除了满足公交公司的虚荣心，没人真的需要这些。而且它们完全看不清：可视角度有限，文字更小，还塞进了一堆没必要的花哨内容，只因为实现者做得到。要小心你追求的东西。
afandian查看英文原文
In Oxford city centre, and probably elsewhere, they replaced the live signage at bus stops. These tell you when the next bus is due, the time and date, etc. They used to be amber LED matrixes. High contrast, high legibility, low resolution. Now they are colour LCD panels. High resolution, high tech, probably high cost. Now they can show the logos of the bus operators, smooth scrolling text and any arbitrary images. None of which anyone really needs except perhaps the egos of the bus companies. And they are completely illegible. The viewing angle is limited, the text is smaller, unnecessary stuff is there because the implementors can. Careful what you ask for.
质疑高分辨率显示不只是屏贵，还会抬高处理器、存储、PCB 和软件成本；作者低估了替代方案的总成本。
作者的很多观点值得考虑（尤其是七段管除数字外别乱用这一点），但拜托：原文大意说，某定制显示面板本可以做得更安全、更易读、更通用、更便宜，如果改用高分辨率显示的话。显然作者从没真正用两种方案做过产品。显示器很贵，而且不只是面板本身贵。通常处理器要升一两级：七段管或点阵可由 4 位单片机驱动；VGA 分辨率显示则需要高端单片机，如今一般是 32 位 Cortex-M3/M4；真正的“高分辨率”屏会把你推到应用处理器（i.MX）级别。听起来不多，但这几档连同外围电路的典型成本大约是 1 美元、10 美元、100 美元。因为小单片机几乎不需要外围；而高端应用处理器需要 DDR SDRAM、NAND flash、PMIC，很可能还是 BGA，要在 8 层板上布线，周围跑 GHz 级 MIPI 信号，更别提软件复杂度了……这些都能做，如果确实让产品更好就该做；但如果产品没收益，就别付这个成本。
exmadscientist查看英文原文
Many of the author's points are worth consideration (especially regarding use of 7-segment displays for anything other than digits), but come on: > A bespoke display panel was used; itwould have been safer, easier to read, more versatile, andcheaper had a hi-res display been used Obviously the author has never actually tried to design a product both ways. Displays are expensive, and I'm not just talking about the panel itself. Usually you must go up a level or two on your processor hierarchy: a 7-segment display or matrix can be driven by a 4-bit microcontroller. A VGA-resolution display will require a high-end microcontroller, usually a 32-bit Cortex-M3 or M4 these days. A true "hi-res" display, which has admittedly drifted up in quality since the author wrote this, will push you into application processor (i.MX) territory. That may not sound like much, but typical costs for those levels including supporting circuitry are in the neighborhood of $1, $10, and $100. This is because the tiny microcontroller needs nothing; the fancy apps processor needs DDR SDRAM, NAND flash, a PMIC, and is probably a BGA you have to route on an 8-layer board, with GHz MIPI signalling flying around, and don't even get me started on the software complexity.... You can do all of these things, and if it genuinely makes your product better, you should. But do not pay the cost if your product does not benefit. A coworker of mine likes to tell a story about cutting the sales price of a medical device from $5,000 to $500. The main source of that savings would have been removing the display. (Yes, really. Medical displays are expensive.) This was rejected by product management, on the reason that "The people who buy this stuff are not the people who use it. The people who buy it like shiny-looking things. It is much easier to sell a product that looks complicated and expensive to a procurement officer than one that looks simple and possibly less advanced, even if they are the same inside. That display does do something -- it sells units." In operation, the display had a bunch of whiz-bang crap that could have been deleted with no consequence, and two touch buttons for "arm" and "stop". Of course, you had to use a physical button to actually activate it; IEC 60601 wouldn't allow using just a touchscreen to start or stop a medical procedure! So the display really didn't do anything functional.... So, no, high-res displays are not always the answer. Think carefully about the cost!
亲历酒厂真实误配置造成七位数损失；七段管不保证失效安全，结论应是别用在关键位置。
这篇论文真奇特。我在一家大型酒厂工作过，那里有几十个储酒罐，每个最大到 10 万升。单个罐出问题就可能是数百万美元损失。我们那台极其昂贵的微氧泵用的是一个标准七段数码管，旁边配一排小红色 LED 表示开/关，再旁边是罐号列表。七段管用来读取和设置单位（μg、mg 等）、量和氧气输送速率。我从没犯过配置错误，但一位资深酿酒师犯过，结果导致客户愤怒、七位数损失和许多合同丢失。我因此有两个看法：七段管通常不是失效安全的，还可能“失效致命”（图 1）。而我这辈子用过的带七段读数的工具（手持或其他）从没实际失效过，我也从没因歧义读错过。所以，我认为更好的结论是：不要在重要场合使用七段数码管。
malwarebytess查看英文原文
What a quirky paper. I worked at a major winery that had dozens of tanks, up to 100,000 liters each. A single failure in a tank represented many millions of dollars. Our extremely expensive micro-oxygen pump used a standard 7-segment display combined with little red LEDs representing On/Off next to a list of tank numbers. The 7-segment display was used to read and to set the units (μg, mg , etc), amount, and rate of oxygen delivery. I never made the mistake of misconfiguring, but one of the senior Winemakers did resulting in some very angry customers, a 7 figure loss, and many contracts. I'm left with two opinions: 7-segment displays are not usually fail-safe, and have the potential to be fail-deadly (figure 1). In my entire life I've never had occasion to use a tool with a 7-segment readout, handheld or otherwise, that actually failed or that I misread due to ambiguity. So, I think a better conclusion would be don't use 7-segment displays in important places.
补充小批量/业余项目用带屏 ESP32 等“性能过剩”方案更省事，并期待廉价通用工业板减少定制开发。
对业余爱好者用途（通常只做 1 件，运费和工时比 1 美元的元件成本更重要），我发现“性能过剩”通常值得。带显示屏的 ESP32 开发板约 10 美元，通用、易用，而且规模效应让它们便宜。还不够？把手机粘上去。我很惊讶没有一种常见的“性能过剩”工业板，类似非常便宜的智能手机，去掉电池/充电、摄像头和 GPS，加上各种协议/引脚接口以连接其他设备。便宜智能手机不到 30 美元，所以这种板子应该不会很贵，还能在为需要触摸屏和 WiFi 的设备（比如高端洗衣机或冰箱）做定制硬件时省下大量开发时间和成本。
tgsovlerkhgsel查看英文原文
For hobbyist use (where quantity is usually 1 and shipping and work matter more than a $1 part cost), I've found that going overkill is usually worth it. A ESP32 board with a built in display costs around $10, and is pretty universal and easy to use, and scale makes them cheap. Need more? Stick a phone to it. I'm surprised there isn't some common "overkill" industrial board similar to a very cheap smartphone, minus battery/charging, camera, and GPS, plus connectors for various protocols/pins for interfacing with other devices. Cheap smartphones cost less than $30, so such a board shouldn't the very expensive, and it could save a lot of development time and costs for bespoke hardware design when building devices that need a touchscreen and WiFi (think a fancy washing machine or fridge).
相关资源
从这篇论文所有 HN 讨论帖的全部评论里整理出来的链接。
历次讨论
- 2021Hacker News 讨论帖109 分 · 78 评
- 2024Hacker News 讨论帖117 分 · 130 评
同领域还值得读
像素不是小方块 (1995)
A Pixel Is Not A Little Square (1995)
这篇 1995 年的经典备忘录讨论数字图像中像素应如何理解：像素不是有面积的小方块，而是对连续图像的采样点，重建连续图像需要合适的重建滤波器。
图形学与人机交互2014 · 372 分 · 241 评进阶 · 约 20 分钟编辑精选自动化的讽刺（Ironies of Automation，1983）
Ironies of Automation (1983)
这篇 1983 年的经典论文讨论自动化（尤其过程控制）中的悖论：自动化越先进，人类操作员在异常处理和人工接管中的作用可能越关键；自动化常把常规任务拿走，留下监控、诊断和罕见故障处…
图形学与人机交互2019 · 147 分 · 38 评入门 · 约 30 分钟编辑精选GDPR 同意框中的暗黑模式
Dark patterns in GDPR consent boxes
研究 GDPR 同意框中的暗黑模式，即网站如何通过界面设计引导用户同意 Cookie 与追踪。
图形学与人机交互2020 · 328 分 · 122 评 · 引用 6进阶 · 约 30 分钟用稀疏条带在 CPU 上实现高性能 2D 图形渲染（硕士论文）
High-performance 2D graphics rendering on the CPU using sparse strips
这篇硕士论文研究如何在 CPU 上实现高性能的 2D 图形渲染，核心思路是用“稀疏条带（sparse strips）”结构来表示和光栅化路径，以降低内存占用并提升渲染速度。
图形学与人机交互2025 · 281 分 · 35 评专业 · 约 90 分钟GDPR 后的暗模式：同意弹窗及其影响
Dark patterns after the GDPR: consent pop-ups and their influence
研究 GDPR 生效后网站 consent pop-ups 中的 dark patterns，以及这些弹窗如何影响用户的同意选择。
图形学与人机交互2021 · 251 分 · 228 评进阶 · 约 60 分钟