计算机限速
A Speed Limit for Computers

原始链接: https://caolan.uk/notes/2026-07-02_a_speed_limit_for_computers.cm

卡奥兰(Caolan)借鉴伊万·伊里奇(Ivan Illich)1973年的文章《能源与公平》(*Energy and Equity*),探讨了技术的“边际负效用”:即一旦某种力量超过特定阈值,便开始侵蚀社会公平与个人自主。正如无限制的车速会造成系统性不平等与时间匮乏,计算能力的指数级增长也可能正在加剧与现代大型科技公司相关的问题。 作者为计算领域提出了“电动自行车”式的概念——通过对时钟频率或存储空间等指标施加法律与政策限制,来定义一类机器。当前的科技监管侧重于数据问责,这往往加重了小型参与者的负担,却未能制约主导系统的权力。通过建立一种“思维自行车”——即优先考虑交互性与可及性,而非单纯追求性能的受限计算类别——我们或许能够重拾社区基础设施,并培育一种更公平、“低能耗”的现代生活。现在的挑战在于如何界定具体的限制条件,从而让此类机器得以蓬勃发展,并有效地将技术进步的益处与过度集权的危害剥离开来。

Hacker News 上题为“计算机的速度限制”的讨论,围绕着一项关于对计算性能施加限制的争议性提案展开,引发了用户间的激烈辩论。 该观点的批评者认为,所谓的“速度限制”是一个荒谬且不切实际的概念,侵犯了基本权利。他们主张,道路上的物理限速是为了防止伤亡,而对计算施加类似的限制既缺乏合理依据,也未能考虑到数字技术的独特本质。 支持者或愿意探讨该概念的人则认为,不受约束的计算可能会产生显著的负面外部效应,如能源消耗和环境影响,这值得讨论。他们认为,计算速度并非总是固有益处的,我们应该审视极端性能带来的社会成本。 总的来说,这场对话凸显了一个根本性的分歧:一方将计算速度视为至关重要、能改善生活的公共事业,认为绝不应受到限制;而另一方则在探讨大规模计算所带来的生态或社会后果,是否足以证明采取新形式监管的必要性。
相关文章

原文
A Speed Limit for Computers · Caolan

Thursday, 02 July 2026

The home and personal computing revolutions of the 70's, 80's, and 90's put the power of computing in the hands of the masses. For those of us that grew up with these machines, it can be hard to reconcile the computer culture of our youth with the industry we find ourselves in today. But at some point, a threshold was crossed and increasing computing power no longer translated to increased autonomy. Quite the opposite.

In 1973, before the first commercially successful home computer, and with the Western world in the beginnings of an energy crisis, Ivan Illich wrote an essay titled 'Energy and Equity'. In it, he argued that - above a certain threshold - energy had marginal disutility. And that applying more of it would yield negative returns.

What is generally overlooked is that energy and equity can grow concurrently only to a point. Below a threshold of per capita wattage, motors improve the conditions for social progress. Above this threshold, energy grows at the expense of equity. Further energy affluence then means decreased distribution of control over that energy.

To make his argument concrete, he applied this theory to traffic and speed:

Beyond a certain velocity, motorized vehicles create remoteness which they alone can shrink. They create distances for all and shrink them only for a few.

And he singles out speed, not the form of conveyance, as the factor driving transport inequality.

Accelerating speed inevitably concentrates horsepower under the seats of a few and compounds the increasing time-lack of most commuters with the further sense that they are lagging behind.

Eventually, in 1974, the UK - feeling the pinch of the oil crisis - did introduce restrictions on vehicle speed to more equitably ration fuel.

Slowly and reluctantly, we have continued to constrict vehicle speed ever since.

But while we still fall short of the "low-energy, convivial modernity" imagined by Ivan Illich, politically determined limits on what a vehicle is technically capable of are accepted as necessary.

In computing, however, we are at the very beginning of this legal and political journey. Restrictions on the industry tend to focus on data processing and accountability rather than limiting the power of our machines. But imagine the effect of a maximum hard drive size, CPU clock frequency, or network speed.

If we still had the machines of the 80's or 90's, would we face the same problems we have today? Could you build Meta or Anthropic on the backs of Amigas and Acorn Archimedes? If not, technological capability plays a role. So how many of our problems derive from the inherent inequity of increased computing power?

Of course, there's no sense dragging everyone back to the 70's to re-run the home computing revolution. But I'd like to highlight a technology in a different category that is defined by its restraint: the e-bike.

In the UK and Europe, the motor of the 'electrically assisted pedal cycle' (EPAC) cuts out at 15 mph [25 km/hour]. This is not a technological limitation but a legal and political choice that slots it neatly into the infrastructure built for the regular human-powered bicycle. That rule might have been a convenience, but I think, surprisingly, it's now the e-bike's most radical intervention.

The speed is what's important. Because, when Ivan Illich attempted to determine a threshold for the disutility for speed, he observed the following:

From our limited information it appears that everywhere in the world, after some vehicle broke the speed barrier of 15 mph [25 km/hour], time scarcity related to traffic began to grow.

In other words, he believes we maximise utility for all by travelling at the speed of a bicycle. And because it is the speed and not the effort of your legs that is crucial, the e-bike increases access to transport (for those who cannot pedal unassisted) without increasing the 'time-lack' of others. It is a new application of energy capable of furthering social progress.

The pedal assisted e-bike is also exempted from many of the restrictions we place on motorbikes and cars thanks to its low speed. It is a whole class of product defined by its energy restraint. And, in the UK, you do not need a license or mandatory insurance to ride one.

In computing, however, we have no comparable machine. Where governments have attempted to apply legal and political constraints to technology - in an attempt to stymie our loss of autonomy to big tech - the collateral damage has been the loss of real community infrastructure that cannot meet their demands.

So, could we make space for a restrained class of computer? A category of machine designed to attract fewer laws and offer some "low-energy, convivial modernity" by its use?

What limits would it have, this bicycle for the mind?

联系我们 contact @ memedata.com