Research you can reproduce
We compare technical options under the same workload and publish everything needed to check the result: the test bench code, commands, versions and raw results. No measurement, no publication.
Publications
Internal engineering experiments
Besides public measurements, InterLAB has internal experimental tools for decision analysis and scenario modelling. We use them to test approaches before offering them in projects.
These are prototypes for internal use, not products or services: they aren't sold, haven't been externally validated, and we don't cite results from them. Their names, design, algorithms and data aren't published. If an approach passes the protocol below, it will appear here as a separate publication — with its method and limitations.
Publication protocol
- 01
Question and hypothesis before measuring
We write down what is compared, which metrics count and what a fair comparison means.
- 02
Same conditions
Options run on the same machine, data and configuration, with several repetitions; background load is noted.
- 03
Raw data and bench code
Every publication comes with the script, run commands, versions and unprocessed results, so you can get your own numbers.
- 04
Limitations stated
We say what the result does not cover: other hardware, network, data volumes, versions. One machine is not industry statistics.
How to cite and check
URL and date
Give the publication URL and date. Each study ends with ready-made citation text and the list of files.
Numbers with their conditions
A median without the version, data size and environment is misleading. Keep the measurement conditions next to the number.
Check on your own machine
The bench code, data generation and commands are open: rerun the measurement and compare — the data checksum shows the dataset is the same.
Errors and corrections
Found a mistake in the method, data or conclusions? Write to hello@interlab.app with the study number.
What you won’t find here
- Numbers without a method and raw data
- Client data or vulnerability details without permission
- “Benchmarks” run against other people’s production systems
- Conclusions broader than the measurement allows
Need to measure something in your product?
Load, page speed or database queries can be measured on your own environment as part of an audit or a consultation.
- The first short conversation is free — we choose the format
- NDA — yours or ours
- Pricing after a brief and an agreed scope