Well, we found data at the Internet Archive: Wayback Machine.

http://archive.org/web/web.php

https://en.wikipedia.org/wiki/Wayback_Machine

Here's our take on what we dug up.

We found text documents of the GLOBAL Temperature Anomalies or GLB.Ts.txt for at least one different date in the past, 2005-08, and compared to the current date 2018-01 for the range from 1880-01 to 2005-08.

What was interesting is even though there is an established anomaly base period of 1951 to 1980 of 14 °C for the data to deviate from, the presented data each year changes a little bit at the front and back end to give the impression the most recent anomaly data is increasing, more and more, each time the data is released for temperature readings that have already been established.

To put in analogy, it's like taking a temperature reading today then taking another tomorrow and the reading you took the day before increased even tough it is extremely improbable that a reading today affects the reading of yesterday... it just doesn't happen... ever.

Presenting data in anomaly fashion should not affect the past either; since the established based period is also in the past, 1951-1980.

No matter how this is done, there is one plausible reason for the change, deliberate manipulation to influence perception.

You can look at the raw data we obtained here: ftp://www.jadexcode.com/GLB.Ts-00

More specifically:

ftp://www.jadexcode.com/GLB.Ts-00/GLB.Ts-2005-08.txt

and

ftp://www.jadexcode.com/GLB.Ts-00/GLB.Ts-2018-01.txt

or download our complied and converted data in Excel format at:

ftp://www.jadexcode.com/GLB.Ts-00/GLB.Ts-CompliedTemperatureDataTablesReformatted-0.xlsm

Essentially we took out the frog boiled data and reset it to a close approximation of the actual... ta da.